**Modelling, Simulation Methods for Intelligent Transportation Systems**

George Papageorgiou and Athanasios Maimaris *European University Cyprus and University of Cyprus Cyprus* 

## **1. Introduction**

100 Intelligent Transportation Systems

W. F. A. EI-Washed (2001), A multi-objective transportation problem under fuzziness, *Fuzzy* 

Y. P. Aneja and K. P. K. Nair (1979), Bicriteria transportation problems, *Management Science*,

*Sets and Systems*, Vol. 117, pp. 27–33.

Vol. 25, pp. 73–78.

Effective transportation systems lead to the efficient movement of goods and people, which significantly contribute to the quality of life in every society. In the heart of every economic and social development, there is always a transportation system. Meanwhile, traffic congestion has been increasing worldwide because of increased motorization, urbanization, population growth, and changes in population density. This threatens the social and economic prosperity of communities all over the world. Congestion reduces utilization of the transportation infrastructure and increases travel time, air pollution, and fuel consumption. Therefore, managing and controlling transportation systems becomes a high priority task for every community, as it constitutes a matter of survival and prosperity for humanity.

In the search for meeting the demand for more traffic capacity, it has been realised repeatedly that building more roads is no longer a feasible solution due to the high cost and/or scarcity of land especially in metropolitan areas. In addition, the length of time that it takes to build additional roads and the disruption that this introduces to the rest of the traffic network makes the option of building new roads as the worst case scenario. The current highway transportation system runs almost open loop whereas traffic lights at surface streets are still lacking the intelligence that is necessary to reduce delays and speed up traffic flows. The recent advances in electronics, communications, controls, computers, and sensors provide an opportunity to develop appropriate transportation management policies and strategies in order to effectively utilize the existing infrastructure rather than building new road systems. The use of technologies will help provide accurate traffic data, implement control actions, and in general reduce the level of uncertainty and randomness that exists in today's transportation networks. The successful implementation of intelligent transportation systems will require a good understanding of the dynamics of traffic on a local as well as global system level and the effect of associated phenomena and disturbances such as shock wave generation and propagation, congestion initiation and so on. In addition, the understanding of human interaction within the transportation system is also crucial.

Transportation systems and traffic phenomena constitute highly complex dynamical problems where simplified mathematical models are not adequate for their analysis. There

Modelling, Simulation Methods for Intelligent Transportation Systems 103

Human factor modeling (Koppa, 1999), deals with salient performance aspects of the human element in the context of the human-machine interactive system. These include perceptionreaction time, control movement time, responses to: traffic control devices, movement of other vehicles, hazards in the roadway, and how different segments of the population differ in performance. Further, human factors theory deals with the kind of control performance that underlies steering, braking, and speed control. Human factors theory provides the basis for the development of car following models. Car following models (Rothery, 1992), examine the manner in which individual vehicles (and their drivers) follow one another. In general, they are developed from a stimulus-response relationship, where the response of successive drivers in the traffic stream is to accelerate or decelerate in proportion to the magnitude of the stimulus. Car following models recognize that traffic is made up of discrete particles or driver-vehicle units and it is the interactions between these units that determine driver behavior, which affects speed-flow-density patterns. On the other hand, continuum models (Kuhne & Michalopoulos, 1997) are concerned more with the overall statistical behavior of the traffic stream rather than with the interactions between the particles. Following the continuum model paradigm, macroscopic flow models (J. C. Williams, 1997), discard the microscopic view of traffic in terms of individual vehicles or individual system components (such as links or intersections) and adopt instead a macroscopic view of traffic in a network. Macroscopic flow models consider variables such as flow rate, speed of flow, density and ignore individual responses of vehicles. Traffic impact models (Ardekani, Hauer, & Jamei, 1992) deal with traffic safety, fuel consumption and air quality models. Traffic safety models describe the relationship between traffic flow and accident frequency. Unsignalized intersection theory (Troutbeck & Brilon, 1997) deals with gap acceptance theory and the headway distributions used in gap acceptance calculations. Traffic flow at signalized intersections (Rouphail, Tarko, & Li, 1996) deals with the statistical theory of traffic flow, in order to provide estimates of delays and queues at isolated intersections, including the effect of upstream traffic signals. Traffic simulation modeling (Lieberman & Rathi, 1996) deals with the traffic models that are embedded in simulation packages and the procedures that are being used for conducting simulation

Mathematically the problem of modelling vehicle traffic flow can be solved at two main observation scales: the microscopic and the macroscopic levels. In the microscopic level, every vehicle is considered individually, and therefore for every vehicle we have an equation that is usually an ordinary differential equation (ODE). At a macroscopic level, we use the analogy of fluid dynamics models, where we have a system of partial differential equations, which involves variables such density, speed, and flow rate of traffic stream with

The microscopic model involves separate units with characteristics such as speed, acceleration, and individual driver-vehicle interaction. Microscopic models may be classified in different types based on the so-called car-following model approach, as it will be discussed in section 4. The car-following modelling approach implies that the driver adjusts his or her acceleration according to the conditions of leading vehicles. In these models, the vehicle position is treated as a continuous function and each vehicle is governed by an ODE that depends on speed and distance of the car in the front. Another type of microscopic model involve the use of Cellular Automata or vehicle hopping models which

experiments.

respect to time and space.

is a need for more advanced methods and models in order to analyse the causality, coupling, feedback loops, and chaotic behaviour involved in transportation problem situations. Traffic modelling can facilitate the effective design and control of today's complex transportation systems. Mathematical models cannot always accurately capture the high complexity and dynamicity of traffic systems. For this reason computer simulation models are developed and tuned to describe the traffic flow characteristics on a given traffic network. Once a computer simulation model is developed and validated using real data, different scenarios and new control strategies can be developed and simulated and evaluated before proposed for an actual implementation.

This chapter presents an overview of traffic flow modelling at the microscopic and macroscopic levels, a review of current traffic simulation software, as well as several methods for managing and controlling the various transportation system modes. In particular, section 2, examines the field of traffic flow theory and the concept of macroscopic vs. microscopic ways of modelling transportation systems. The derivation of traffic flow theory based on the law of conservation of mass, and the relationships between flow speed and density are presented in section 3 under the topic of macroscopic models. Section 4 analyses microscopic car following models and discusses advantages and limitations. Section 5 reviews various some of the most sophisticated traffic software modelling tools, all in relation to intelligent transportation systems. Finally, a summary of recent intelligent transportation systems studies carried out by the authors is provided in section 6 and conclusions are drawn in section 7.

## **2. Traffic flow modelling**

The study of traffic flow (May, 1990), and in particular vehicular traffic flow, is carried out with the aim of understanding and assisting in the prevention and remedy of traffic congestion problems. The first attempts to develop a mathematical theory for traffic flow date back to the 1930s (Adams, 1937; Greenshields, 1935a), but despite the continuous research activity in the area we do not have yet a satisfactory mathematical theory to describe real traffic flow conditions. This is because traffic phenomena are complex and nonlinear, depending on the interactions of a large number of vehicles. Moreover, vehicles do not interact simply by following the laws of physics, but are also influenced by the psychological reactions of human drivers. As a result we observe chaotic phenomena such as cluster formation and backward propagating shockwaves of vehicle speed/density (Bose & Ioannou, 2000) that are difficult if at all possible to be accurately described with mathematical models. According to a state of the art report of the Transportation Research Board (Gartner, Messer, & Rathi, 2001), mathematical models for traffic flow may be classified as: Traffic Stream Characteristics Models, Human Factor Models, Car Following Models, Continuum Flow Models, Macroscopic Flow Models, Traffic Impact Models, Unsignalized Intersection Models, Signalized Intersection Models and Traffic Simulation Models. Below we describe briefly each of the above categories.

Traffic stream characteristics (Hall, 1996) theory involves various mathematical models, which have been developed to characterize the relationships among the traffic stream variables of speed, flow, and concentration or density.

experiments.

102 Intelligent Transportation Systems

is a need for more advanced methods and models in order to analyse the causality, coupling, feedback loops, and chaotic behaviour involved in transportation problem situations. Traffic modelling can facilitate the effective design and control of today's complex transportation systems. Mathematical models cannot always accurately capture the high complexity and dynamicity of traffic systems. For this reason computer simulation models are developed and tuned to describe the traffic flow characteristics on a given traffic network. Once a computer simulation model is developed and validated using real data, different scenarios and new control strategies can be developed and simulated and

This chapter presents an overview of traffic flow modelling at the microscopic and macroscopic levels, a review of current traffic simulation software, as well as several methods for managing and controlling the various transportation system modes. In particular, section 2, examines the field of traffic flow theory and the concept of macroscopic vs. microscopic ways of modelling transportation systems. The derivation of traffic flow theory based on the law of conservation of mass, and the relationships between flow speed and density are presented in section 3 under the topic of macroscopic models. Section 4 analyses microscopic car following models and discusses advantages and limitations. Section 5 reviews various some of the most sophisticated traffic software modelling tools, all in relation to intelligent transportation systems. Finally, a summary of recent intelligent transportation systems studies carried out by the authors is provided in section 6 and

The study of traffic flow (May, 1990), and in particular vehicular traffic flow, is carried out with the aim of understanding and assisting in the prevention and remedy of traffic congestion problems. The first attempts to develop a mathematical theory for traffic flow date back to the 1930s (Adams, 1937; Greenshields, 1935a), but despite the continuous research activity in the area we do not have yet a satisfactory mathematical theory to describe real traffic flow conditions. This is because traffic phenomena are complex and nonlinear, depending on the interactions of a large number of vehicles. Moreover, vehicles do not interact simply by following the laws of physics, but are also influenced by the psychological reactions of human drivers. As a result we observe chaotic phenomena such as cluster formation and backward propagating shockwaves of vehicle speed/density (Bose & Ioannou, 2000) that are difficult if at all possible to be accurately described with mathematical models. According to a state of the art report of the Transportation Research Board (Gartner, Messer, & Rathi, 2001), mathematical models for traffic flow may be classified as: Traffic Stream Characteristics Models, Human Factor Models, Car Following Models, Continuum Flow Models, Macroscopic Flow Models, Traffic Impact Models, Unsignalized Intersection Models, Signalized Intersection Models and Traffic Simulation

Traffic stream characteristics (Hall, 1996) theory involves various mathematical models, which have been developed to characterize the relationships among the traffic stream

evaluated before proposed for an actual implementation.

Models. Below we describe briefly each of the above categories.

variables of speed, flow, and concentration or density.

conclusions are drawn in section 7.

**2. Traffic flow modelling** 

Human factor modeling (Koppa, 1999), deals with salient performance aspects of the human element in the context of the human-machine interactive system. These include perceptionreaction time, control movement time, responses to: traffic control devices, movement of other vehicles, hazards in the roadway, and how different segments of the population differ in performance. Further, human factors theory deals with the kind of control performance that underlies steering, braking, and speed control. Human factors theory provides the basis for the development of car following models. Car following models (Rothery, 1992), examine the manner in which individual vehicles (and their drivers) follow one another. In general, they are developed from a stimulus-response relationship, where the response of successive drivers in the traffic stream is to accelerate or decelerate in proportion to the magnitude of the stimulus. Car following models recognize that traffic is made up of discrete particles or driver-vehicle units and it is the interactions between these units that determine driver behavior, which affects speed-flow-density patterns. On the other hand, continuum models (Kuhne & Michalopoulos, 1997) are concerned more with the overall statistical behavior of the traffic stream rather than with the interactions between the particles. Following the continuum model paradigm, macroscopic flow models (J. C. Williams, 1997), discard the microscopic view of traffic in terms of individual vehicles or individual system components (such as links or intersections) and adopt instead a macroscopic view of traffic in a network. Macroscopic flow models consider variables such as flow rate, speed of flow, density and ignore individual responses of vehicles. Traffic impact models (Ardekani, Hauer, & Jamei, 1992) deal with traffic safety, fuel consumption and air quality models. Traffic safety models describe the relationship between traffic flow and accident frequency. Unsignalized intersection theory (Troutbeck & Brilon, 1997) deals with gap acceptance theory and the headway distributions used in gap acceptance calculations. Traffic flow at signalized intersections (Rouphail, Tarko, & Li, 1996) deals with the statistical theory of traffic flow, in order to provide estimates of delays and queues at isolated intersections, including the effect of upstream traffic signals. Traffic simulation modeling (Lieberman & Rathi, 1996) deals with the traffic models that are embedded in simulation packages and the procedures that are being used for conducting simulation

Mathematically the problem of modelling vehicle traffic flow can be solved at two main observation scales: the microscopic and the macroscopic levels. In the microscopic level, every vehicle is considered individually, and therefore for every vehicle we have an equation that is usually an ordinary differential equation (ODE). At a macroscopic level, we use the analogy of fluid dynamics models, where we have a system of partial differential equations, which involves variables such density, speed, and flow rate of traffic stream with respect to time and space.

The microscopic model involves separate units with characteristics such as speed, acceleration, and individual driver-vehicle interaction. Microscopic models may be classified in different types based on the so-called car-following model approach, as it will be discussed in section 4. The car-following modelling approach implies that the driver adjusts his or her acceleration according to the conditions of leading vehicles. In these models, the vehicle position is treated as a continuous function and each vehicle is governed by an ODE that depends on speed and distance of the car in the front. Another type of microscopic model involve the use of Cellular Automata or vehicle hopping models which

Modelling, Simulation Methods for Intelligent Transportation Systems 105

the vehicle traffic flow fundamentals for the macroscopic modeling approach. The relationship between density, velocity, and flow is also presented. Then we derive the equation of conservation of vehicles, which is the main governing equation for scalar macroscopic traffic flow models. The macroscopic models for traffic flow, whether they are one-equation or a system of equations, are based on the physical principle of conservation. When physical quantities remain the same during some process, these quantities are said to be conserved. Putting this principle into a mathematical representation, it becomes possible

Drawing an analogy between vehicle dynamics and fluid dynamics (Kuhne & Michalopoulos, 1997) let us consider a unidirectional continuous road section with two counting stations 1,2 at positions 1 *x* and 2 *x* . The spacing between stations is *x* . In such a case, the number of cars in a segment of a highway *x* is a physical quantity, and the process is to keep it fixed so that the number of cars coming in equals the number of cars

As it will be shown there is a close interrelationship between three traffic variables that is density, velocity and traffic flow. Suppose that in the above scenario, cars are moving with

also constant. Let an observer measure the number of cars *N* per unit time *t* that pass

Let *N*1 be the number of cars passing station 1 and *N*2 be the number of cars passing station 2 and *t* the duration of the observer counting time. Let *q* be the flow rate i.e. the

*N qt* ( )

Assuming that *x* is short enough so that vehicle density is uniform, then the increase in

Assuming conservation of vehicles and no sinks or sources exist in the section of the

or *N x*

 or 0 *<sup>q</sup> x t* 

*N x*

 and <sup>2</sup> 2 *N <sup>q</sup> <sup>t</sup>*

or *N N N qt* 2 1

1

1 *N <sup>q</sup> <sup>t</sup>*

*N N* 2 1 *<sup>q</sup> t t* 

> 

 *qt x* 

For a build-up of cars therefore *N* will be negative. Thus

such that the distance *d* between the cars is

to predict the density and velocity patterns at a future time.

number of cars passing a particular station per unit time, then

going out of the segment.

him/her (i.e. the traffic flow *q* ).

density during time *t* is given by

roadway,

constant velocity *v* , and constant density

differ from the car-following approach in that they are fully discrete time models. They consider the road as a string of cells that are either empty or occupied by one vehicle. One such model is the Stochastic Traffic Cellular Automata (Nagel, 1996; Nagel & Schreckenberg, 1992) model. Further, a more recent approach is currently under heavy research with the use of agent based modeling (Naiem, Reda, El-Beltagy, & El-Khodary, 2010).

Microscopic approaches are generally computationally intense, as each car has an ODE to be solved at each time step, and as the number of cars increases, so does the size of the system to be solved. Analytical mathematical microscopic models are difficult to evaluate but a remedy for this is the use of microscopic computer simulation. In such microscopic traffic models, vehicles are treated as discrete driver-vehicle units moving in a computer-simulated environment.

On the other hand, macroscopic models aim at studying traffic flow using a continuum approach, where it is assumed that the movement of individual vehicles exhibit many of the attributes of fluid motion. As a result, vehicle dynamics are treated as fluid dynamics. This idea provides an advantage since detailed interactions are overlooked, and the model's characteristics are shifted toward the more important parameters such as flow rate, concentration, or traffic density, and average speed, all being functions of one-dimensional space and time. This class of models is represented by partial differential equations. Modeling vehicular traffic via macroscopic models is achieved using fluid flow theory in a continuum responding to local or non-local influences. The mathematical details of such models are less than those of the microscopic ones. The drawback of macroscopic modeling is the assumption that traffic flow behaves like fluid flow, which is a rather harsh approximation of reality. Vehicles tend to interact among themselves and are sensitive to local traffic disturbances, phenomena that are not captured by macroscopic models. On the other hand, macroscopic models are suitable for studying large-scale problems and are computationally less intense especially after approximating the partial differential equation with a discrete time finite order equation.

There exists also a third level of analysis the so called mesoscopic level, which is somewhere between the microscopic and the macroscopic levels. In a mesoscopic or kinetic scale, which is an intermediate level, we define a function *f* (, , ) *txv* , which expresses the probability of having a vehicle at time *t* in position *x* at velocity *v* . This function, following methods of statistical mechanics, can be computed by solving an integro-differential equation, like the Boltzmann Equation (K. Waldeer, 2006; K. T. Waldeer, 2004).

The choice of the appropriate model depends on the level of detail required and the computing power available. Because of advancements in computer technology in recent years, the trend today is towards utilizing microscopic scale mathematical models, which incorporate human factors and car following models as a driver-vehicle behavior unit.

In the next two sections, macroscopic and microscopic models are examined in more detail.

#### **2.1 Macroscopic traffic flow models**

Macroscopic flow models (J. Williams, 1996), discard the real view of traffic in terms of individual vehicles or individual system components such as links or intersections and adopt instead a macroscopic fluid view of traffic in a network. In this section, we will cover

differ from the car-following approach in that they are fully discrete time models. They consider the road as a string of cells that are either empty or occupied by one vehicle. One such model is the Stochastic Traffic Cellular Automata (Nagel, 1996; Nagel & Schreckenberg, 1992) model. Further, a more recent approach is currently under heavy research with the use

Microscopic approaches are generally computationally intense, as each car has an ODE to be solved at each time step, and as the number of cars increases, so does the size of the system to be solved. Analytical mathematical microscopic models are difficult to evaluate but a remedy for this is the use of microscopic computer simulation. In such microscopic traffic models, vehicles are treated as discrete driver-vehicle units moving in a computer-simulated

On the other hand, macroscopic models aim at studying traffic flow using a continuum approach, where it is assumed that the movement of individual vehicles exhibit many of the attributes of fluid motion. As a result, vehicle dynamics are treated as fluid dynamics. This idea provides an advantage since detailed interactions are overlooked, and the model's characteristics are shifted toward the more important parameters such as flow rate, concentration, or traffic density, and average speed, all being functions of one-dimensional space and time. This class of models is represented by partial differential equations. Modeling vehicular traffic via macroscopic models is achieved using fluid flow theory in a continuum responding to local or non-local influences. The mathematical details of such models are less than those of the microscopic ones. The drawback of macroscopic modeling is the assumption that traffic flow behaves like fluid flow, which is a rather harsh approximation of reality. Vehicles tend to interact among themselves and are sensitive to local traffic disturbances, phenomena that are not captured by macroscopic models. On the other hand, macroscopic models are suitable for studying large-scale problems and are computationally less intense especially after approximating the partial differential equation

There exists also a third level of analysis the so called mesoscopic level, which is somewhere between the microscopic and the macroscopic levels. In a mesoscopic or kinetic scale, which is an intermediate level, we define a function *f* (, , ) *txv* , which expresses the probability of having a vehicle at time *t* in position *x* at velocity *v* . This function, following methods of statistical mechanics, can be computed by solving an integro-differential equation, like the

The choice of the appropriate model depends on the level of detail required and the computing power available. Because of advancements in computer technology in recent years, the trend today is towards utilizing microscopic scale mathematical models, which incorporate human factors and car following models as a driver-vehicle behavior unit.

In the next two sections, macroscopic and microscopic models are examined in more detail.

Macroscopic flow models (J. Williams, 1996), discard the real view of traffic in terms of individual vehicles or individual system components such as links or intersections and adopt instead a macroscopic fluid view of traffic in a network. In this section, we will cover

of agent based modeling (Naiem, Reda, El-Beltagy, & El-Khodary, 2010).

environment.

with a discrete time finite order equation.

**2.1 Macroscopic traffic flow models** 

Boltzmann Equation (K. Waldeer, 2006; K. T. Waldeer, 2004).

the vehicle traffic flow fundamentals for the macroscopic modeling approach. The relationship between density, velocity, and flow is also presented. Then we derive the equation of conservation of vehicles, which is the main governing equation for scalar macroscopic traffic flow models. The macroscopic models for traffic flow, whether they are one-equation or a system of equations, are based on the physical principle of conservation. When physical quantities remain the same during some process, these quantities are said to be conserved. Putting this principle into a mathematical representation, it becomes possible to predict the density and velocity patterns at a future time.

Drawing an analogy between vehicle dynamics and fluid dynamics (Kuhne & Michalopoulos, 1997) let us consider a unidirectional continuous road section with two counting stations 1,2 at positions 1 *x* and 2 *x* . The spacing between stations is *x* . In such a case, the number of cars in a segment of a highway *x* is a physical quantity, and the process is to keep it fixed so that the number of cars coming in equals the number of cars going out of the segment.

As it will be shown there is a close interrelationship between three traffic variables that is density, velocity and traffic flow. Suppose that in the above scenario, cars are moving with constant velocity *v* , and constant density such that the distance *d* between the cars is also constant. Let an observer measure the number of cars *N* per unit time *t* that pass him/her (i.e. the traffic flow *q* ).

Let *N*1 be the number of cars passing station 1 and *N*2 be the number of cars passing station 2 and *t* the duration of the observer counting time. Let *q* be the flow rate i.e. the number of cars passing a particular station per unit time, then

$$q\_1 = \frac{N\_1}{\Delta t} \text{ and } q\_2 = \frac{N\_2}{\Delta t}$$

$$\Delta q = \frac{N\_2}{\Delta t} - \frac{N\_1}{\Delta t} \text{ or } \Delta N = N\_2 - N\_1 = \Delta q \Delta t$$

For a build-up of cars therefore *N* will be negative. Thus

$$
\Delta \mathbf{N} = (-\Delta q)\Delta t
$$

Assuming that *x* is short enough so that vehicle density is uniform, then the increase in density during time *t* is given by

$$
\Delta \rho = \frac{-\Delta N}{\Delta \mathbf{x}} \text{ or } -\Delta N = \Delta \rho \Delta \mathbf{x}
$$

Assuming conservation of vehicles and no sinks or sources exist in the section of the roadway,

$$-\Delta q \Delta t = \Delta \rho \Delta \chi \quad \text{or} \quad \frac{\Delta q}{\Delta \chi} + \frac{\Delta \rho}{\Delta t} = 0$$

Modelling, Simulation Methods for Intelligent Transportation Systems 107

Based on the above empirical relationship Greenshields derived the following parabolic

<sup>2</sup> ( ) *jam f q vv v* 

Real traffic data reflects somewhat the flow-density relationship for Greenshild's model, which follows a parabolic shape and shows that the flow increases to a maximum which

third relationship that can be drawn is the speed flow function, which is again of parabolic

Following Greenshields steps, Greenberg (Greenberg, 1959) developed a model of speeddensity showing a logarithmic relationship. In Greenberg's model the speed-density

> ln( ) *<sup>f</sup> jam*

*<sup>f</sup> v ve*

Another single regime model is the Underwood model (Underwood, 1961), where the

(Edie, 1961) proposed a multi-regime model which basically combines Greenberg's and

163.9 *v e* 54.9

*v* 28.6ln( )

(Drake, Schofer, & May Jr, 1967) investigated seven speed-density models, including the above, through an empirical test in 1967. According to Drake et al the Eddie formulation gave the best estimates of the fundamental parameters but its root mean square error (rms) was the second lowest. The general conclusion though was that none of the models

More recently, prominent researchers such as (Payne, 1979), (M. Papageorgiou, Blosseville, & Hadj-Salem, 1989) and (Michalopoulos, Yi, & Lyrintzis, 1992) developed macroscopic traffic flow simulation models based on a space-time discretization of the conservation equation. Even though these models are capable of describing complicated traffic

investigated provided a particularly good fit or explanation of the traffic data tested.

*jam*

162.5

*v v*

and then it goes back to zero at high values of density. A

equation for the flow-speed-density relationship,

velocity-density function is represented as follows.

Underwood's model. Eddie suggested that

occurs at some average density

shape.

function is given by

for densities 50 

and for densities 50

Finally, assuming continuity of the medium and infinitesimal increments we get the conservation or continuity equation

$$\frac{\partial q}{\partial \mathbf{x}} + \frac{\partial \rho}{\partial t} = 0$$

In order to solve the above equation we assume that *v f* ( ) and *q v*

$$\frac{\partial}{\partial \mathbf{x}} (\rho v) + \frac{\partial \rho}{\partial t} = 0 \quad \text{or} \quad \frac{\partial}{\partial \mathbf{x}} (\rho f(\rho)) + \frac{\partial \rho}{\partial t} = 0$$

Differentiating with respect to *x* ,

$$\frac{\partial \rho}{\partial \mathbf{x}} f(\rho) + \frac{\partial f(\rho)}{\partial \mathbf{x}} \rho + \frac{\partial \rho}{\partial t} = 0 \text{ or } \frac{\partial \rho}{\partial \mathbf{x}} f(\rho) + \frac{df \hat{\partial} \rho}{d\rho \hat{\partial} \mathbf{x}} \rho + \frac{\partial \rho}{\partial t} = 0$$

$$\Rightarrow (f(\rho) + \rho \frac{df}{d\rho}) \frac{\partial \rho}{\partial \mathbf{x}} + \frac{\partial \rho}{\partial t} = 0$$

The above constitutes a first-order partial differential equation, which can be solved by the method of characteristics. A complete formulation and solution of the above equation has been published (Lighthill & Whitham, 1955). If the initial density and the velocity field are known, the above equation can be used to predict future traffic density. This leads us to choose the velocity function for the traffic flow model to be dependent on density and call it*V*( ) . The above equation assumes no generation or dissipation of vehicles. Sources and sinks may be added by including a function *gxt* ( ,) on the RHS of the equation.

Several velocity-density-flow models have been developed through the years and are classified as single-regime or multi-regime models. Single-regime models assume a continuous relationship between velocity, density, and traffic flow while in multi-regime models the relationship is discontinuous depending on the density levels. Some of the most well-known single-regime and multi-regime models include the Greenshields model, Greenberg model, the Underwood model and Eddie's model. These are described as follows.

The Greenshields Model (Greenshields, 1935b) is a simple and widely used model. It is assumed that the velocity is a linearly decreasing function of the traffic flow density, and it is given by

$$
\upsilon = \upsilon\_f \left( 1 - \frac{\rho}{\rho\_{j\text{am}}} \right)
$$

where *<sup>f</sup> v* is the free flow speed and *jam* is the jam density. The above equation represents a monotonically decreasing function with respect to density. For zero density the model allows free flow speed *<sup>f</sup> v* , while for maximum density *jam* we have 100% congestion where the speed is zero and no car is moving. Real traffic data shows that the speed-density relationship is indeed a rather linear negative slope function.

Based on the above empirical relationship Greenshields derived the following parabolic equation for the flow-speed-density relationship,

$$q = -(\frac{\rho\_{jum}}{v\_f})v^2 + \rho v$$

Real traffic data reflects somewhat the flow-density relationship for Greenshild's model, which follows a parabolic shape and shows that the flow increases to a maximum which occurs at some average density and then it goes back to zero at high values of density. A third relationship that can be drawn is the speed flow function, which is again of parabolic shape.

Following Greenshields steps, Greenberg (Greenberg, 1959) developed a model of speeddensity showing a logarithmic relationship. In Greenberg's model the speed-density function is given by

$$\upsilon = \upsilon\_f - \ln(\frac{\rho}{\rho\_{\text{jom}}})$$

Another single regime model is the Underwood model (Underwood, 1961), where the velocity-density function is represented as follows.

$$\upsilon = \upsilon\_f \stackrel{\rho}{e^{-\rho\_{jm}}}$$

(Edie, 1961) proposed a multi-regime model which basically combines Greenberg's and Underwood's model. Eddie suggested that

for densities 50 

106 Intelligent Transportation Systems

Finally, assuming continuity of the medium and infinitesimal increments we get the

<sup>0</sup> *<sup>q</sup> x t* 

or ( ) <sup>0</sup> *df <sup>f</sup> x dx t*

. The above equation assumes no generation or dissipation of

 

or ( ( )) 0 *<sup>f</sup> x t*

 

 

(() ) 0 *df <sup>f</sup> dxt* 

The above constitutes a first-order partial differential equation, which can be solved by the method of characteristics. A complete formulation and solution of the above equation has been published (Lighthill & Whitham, 1955). If the initial density and the velocity field are known, the above equation can be used to predict future traffic density. This leads us to choose the velocity function for the traffic flow model to be dependent on

vehicles. Sources and sinks may be added by including a function *gxt* ( ,) on the RHS of

Several velocity-density-flow models have been developed through the years and are classified as single-regime or multi-regime models. Single-regime models assume a continuous relationship between velocity, density, and traffic flow while in multi-regime models the relationship is discontinuous depending on the density levels. Some of the most well-known single-regime and multi-regime models include the Greenshields model, Greenberg model,

The Greenshields Model (Greenshields, 1935b) is a simple and widely used model. It is assumed that the velocity is a linearly decreasing function of the traffic flow density, and it

> (1 ) *<sup>f</sup> jam*

*jam* is the jam density. The above equation represents

*jam* we have 100% congestion

*v v*

allows free flow speed *<sup>f</sup> v* , while for maximum density

relationship is indeed a rather linear negative slope function.

a monotonically decreasing function with respect to density. For zero density the model

where the speed is zero and no car is moving. Real traffic data shows that the speed-density

and *q*

  *v*

conservation or continuity equation

Differentiating with respect to *x* ,

density and call it*V*( )

the equation.

is given by

where *<sup>f</sup> v* is the free flow speed and

In order to solve the above equation we assume that *v f* ( )

 

( ) ( ) <sup>0</sup> *<sup>f</sup> <sup>f</sup> x xt*

 

() 0 *v x t*

> 

 

the Underwood model and Eddie's model. These are described as follows.

$$v = 54.9e^{-\frac{\rho}{163.9}}$$

and for densities 50 

$$v = 28.6 \ln(\frac{162.5}{\rho})$$

(Drake, Schofer, & May Jr, 1967) investigated seven speed-density models, including the above, through an empirical test in 1967. According to Drake et al the Eddie formulation gave the best estimates of the fundamental parameters but its root mean square error (rms) was the second lowest. The general conclusion though was that none of the models investigated provided a particularly good fit or explanation of the traffic data tested.

More recently, prominent researchers such as (Payne, 1979), (M. Papageorgiou, Blosseville, & Hadj-Salem, 1989) and (Michalopoulos, Yi, & Lyrintzis, 1992) developed macroscopic traffic flow simulation models based on a space-time discretization of the conservation equation. Even though these models are capable of describing complicated traffic

Modelling, Simulation Methods for Intelligent Transportation Systems 109

Fig. 1. A Generalised Block Diagram of Car Following (adapted from Rothery, 1992).

presented below.

vehicle.

The car-following behaviour is basically, a human interactive process where the driver of the vehicle attempts to reach a stable situation and maintain it by following a leading vehicle, by continuously taking corrective actions like accelerating or decelerating. As it will be seen in the next paragraphs car following models may be classified as Stimulus-Response models, safety distance or collision avoidance models, psychophysical or action point models, and fuzzy logic models. Some of the most widely applied car following models are

Pipes (Pipes, 1967) proposed a theory of car following behaviour based on what he referred to as the "idealized law of separation". The law specifies that each vehicle must maintain a certain prescribed "following distance" from the preceding vehicle. This distance is the sum of a distance proportional to the velocity of the following vehicle and a certain given minimum distance of separation when the vehicles are at rest. Such a model implies that the actions of the following vehicle are only affected by the relative speed between the leading vehicle and the following vehicle. Forbes (Forbes, 1963) modelled car following behaviour by assuming that drivers choose to keep a minimum time gap from the rear end of leading vehicle. The Forbes model of car following also implies that the actions of the following vehicle are only affected by the relative speed between the leading vehicle and the following

The General Motors Research Laboratories published significant amount of work on the carfollowing theory model in a series of papers (Gazis, Herman, & Potts, 1959; Herman, Montroll, Potts, & Rothery, 1959). The basic idea used here is that the actions of the following vehicle in terms of acceleration or deceleration are a function of a single stimulus and the sensitivity of the following vehicle to the stimulus under the prevailing conditions. The stimulus is assumed to be the relative speed between leading and the following vehicle. Sensitivity to the stimulus is assumed to be affected by the distance headway between the

Other approaches such as those by Rockwell et al (Rockwell, Ernest, & Hanken, 1968) present a regression based car-following model, which takes into consideration two leading vehicles, and Chakroborty and Kikuchi (Kikuchi, Chakroborty, & Engineering, 1992) a

Consolidating the various approaches of car following models it can be concluded that the general assumption about the interaction between a leader and follower car is governed by

leading vehicle and the following vehicle as well as the speed of following vehicle.

Fuzzy Inference based car-following model.

the following equation (Rothery, 1992).

phenomena with considerable accuracy, their main limitations arise in their inability to accurately simulate severe traffic congestion situations, where the conservation equation does not represent the traffic flow so well.

On the other hand, it is desirable to use macroscopic models if a good model can be found that satisfactorily describes the traffic flow for the particular traffic problem situation. The advantage of macroscopic models is their flexibility since detailed interactions are overlooked, and the model's characteristics are shifted toward important parameters such as flow rate, concentration or traffic density, and average speed. If the transportation/traffic problem demands more detail and accuracy such as the case of evaluating the effects of closely spaced intersections or bus priority systems on the traffic network then one should resort to microscopic models, which are described in the next section.

#### **2.2 Microscopic driver-vehicle behaviour models**

In this section well known car-following microscopic traffic flow models are presented and evaluated on their capability to realistically model traffic flow at the vehicle level.

In order to derive a one-dimensional simple car following model we first assume that cars do not pass each other. Then the idea is that a car in one-dimension can move and accelerate forward based on two parameters; the headway distance between the current car and the leading car, and their speed difference. Hence, it is called following, since a car from behind follows the one in the front.

Car-following models are based on the assumption that a stimulus response relationship exists that describes the control process of a driver-vehicle unit. This concept is expressed with the stimulus response equation where response is proportionally analogous to a stimulus based on a certain proportionality factor (Rothery, 1992) .

As seen later on in this section the various car following models incorporate a variant of the following stimulus response equation.

$$
\ddot{\boldsymbol{x}}\_f(t) = \mathcal{X} \cdot \frac{\left(\dot{\boldsymbol{x}}\_l(t) - \dot{\boldsymbol{x}}\_f(t)\right)}{\left(\boldsymbol{x}\_l(t) - \boldsymbol{x}\_f(t)\right)}.
$$

where *<sup>f</sup> x* is the one-dimensional position of the following vehicle, *<sup>l</sup> x* is the onedimensional position of the leading vehicle, and *t* is time. The response function here is taken as the acceleration of the following vehicle as the driver experiences inertia forces and has direct control on acceleration/deceleration through the accelerator and brake pedals.

The above stimulus-response equation of car-following is a simplified description of a complex phenomenon. A generalization of car following in a conventional control theory block diagram is shown in figure 1. As seen in figure 1 a more complete representation of car following would include a set of equations that are able to model the dynamical properties of the vehicle and the roadway characteristics. It would also include the psychological and physiological properties of drivers, as well as couplings between vehicles, other than the forward nearest neighbours and other driving tasks such as lateral control, the state of traffic, and emergency conditions and other factors.

phenomena with considerable accuracy, their main limitations arise in their inability to accurately simulate severe traffic congestion situations, where the conservation equation

On the other hand, it is desirable to use macroscopic models if a good model can be found that satisfactorily describes the traffic flow for the particular traffic problem situation. The advantage of macroscopic models is their flexibility since detailed interactions are overlooked, and the model's characteristics are shifted toward important parameters such as flow rate, concentration or traffic density, and average speed. If the transportation/traffic problem demands more detail and accuracy such as the case of evaluating the effects of closely spaced intersections or bus priority systems on the traffic network then one should

In this section well known car-following microscopic traffic flow models are presented and

In order to derive a one-dimensional simple car following model we first assume that cars do not pass each other. Then the idea is that a car in one-dimension can move and accelerate forward based on two parameters; the headway distance between the current car and the leading car, and their speed difference. Hence, it is called following, since a car from behind

Car-following models are based on the assumption that a stimulus response relationship exists that describes the control process of a driver-vehicle unit. This concept is expressed with the stimulus response equation where response is proportionally analogous to a

As seen later on in this section the various car following models incorporate a variant of the

() () ( ) () () *l f*

where *<sup>f</sup> x* is the one-dimensional position of the following vehicle, *<sup>l</sup> x* is the onedimensional position of the leading vehicle, and *t* is time. The response function here is taken as the acceleration of the following vehicle as the driver experiences inertia forces and has direct control on acceleration/deceleration through the accelerator and brake pedals.

The above stimulus-response equation of car-following is a simplified description of a complex phenomenon. A generalization of car following in a conventional control theory block diagram is shown in figure 1. As seen in figure 1 a more complete representation of car following would include a set of equations that are able to model the dynamical properties of the vehicle and the roadway characteristics. It would also include the psychological and physiological properties of drivers, as well as couplings between vehicles, other than the forward nearest neighbours and other driving tasks such as lateral control,

 

*f*

the state of traffic, and emergency conditions and other factors.

*x t*

 

*xt x t*

*l f*

*xt x t*

(Rothery, 1992) .

evaluated on their capability to realistically model traffic flow at the vehicle level.

resort to microscopic models, which are described in the next section.

**2.2 Microscopic driver-vehicle behaviour models** 

stimulus based on a certain proportionality factor

following stimulus response equation.

follows the one in the front.

does not represent the traffic flow so well.

Fig. 1. A Generalised Block Diagram of Car Following (adapted from Rothery, 1992).

The car-following behaviour is basically, a human interactive process where the driver of the vehicle attempts to reach a stable situation and maintain it by following a leading vehicle, by continuously taking corrective actions like accelerating or decelerating. As it will be seen in the next paragraphs car following models may be classified as Stimulus-Response models, safety distance or collision avoidance models, psychophysical or action point models, and fuzzy logic models. Some of the most widely applied car following models are presented below.

Pipes (Pipes, 1967) proposed a theory of car following behaviour based on what he referred to as the "idealized law of separation". The law specifies that each vehicle must maintain a certain prescribed "following distance" from the preceding vehicle. This distance is the sum of a distance proportional to the velocity of the following vehicle and a certain given minimum distance of separation when the vehicles are at rest. Such a model implies that the actions of the following vehicle are only affected by the relative speed between the leading vehicle and the following vehicle. Forbes (Forbes, 1963) modelled car following behaviour by assuming that drivers choose to keep a minimum time gap from the rear end of leading vehicle. The Forbes model of car following also implies that the actions of the following vehicle are only affected by the relative speed between the leading vehicle and the following vehicle.

The General Motors Research Laboratories published significant amount of work on the carfollowing theory model in a series of papers (Gazis, Herman, & Potts, 1959; Herman, Montroll, Potts, & Rothery, 1959). The basic idea used here is that the actions of the following vehicle in terms of acceleration or deceleration are a function of a single stimulus and the sensitivity of the following vehicle to the stimulus under the prevailing conditions. The stimulus is assumed to be the relative speed between leading and the following vehicle. Sensitivity to the stimulus is assumed to be affected by the distance headway between the leading vehicle and the following vehicle as well as the speed of following vehicle.

Other approaches such as those by Rockwell et al (Rockwell, Ernest, & Hanken, 1968) present a regression based car-following model, which takes into consideration two leading vehicles, and Chakroborty and Kikuchi (Kikuchi, Chakroborty, & Engineering, 1992) a Fuzzy Inference based car-following model.

Consolidating the various approaches of car following models it can be concluded that the general assumption about the interaction between a leader and follower car is governed by the following equation (Rothery, 1992).

Modelling, Simulation Methods for Intelligent Transportation Systems 111

where *ll v* the speed of the vehicle in front of the immediate leading vehicle and *ll d* the

Bexelius (Bexelius, 1968) also suggested a car following model that takes into account two

1 2 ( )( ) *<sup>f</sup> ll <sup>f</sup> <sup>l</sup> <sup>f</sup> a Pv v Pv v*

More complex car following models of similar nature were developed by (Wicks, Lieberman, Associates, & Division, 1980) where the NETSIM software is based, and by (Bullen, 1982) where FRESIM is based. Further, (Gipps, 1981) developed a safety distance or behavioural model, which is employed by AIMSUN. (Fritzsche, 1994) and (Wiedemann, 1974) developed the so called action point or phycho-physical models for Paramics and VISSIM respectively. Some more modern approaches to car following models make use of fuzzy logic algorithms (Gonzalez-Rojo, Slama, Lopes, & Mora-Camino, 2002; Yikai, Satoh, Itakura, Honda, & Satoh, 1993). Interesting is also the employment of the System Dynamics modelling principles in car following. More recently, (Mehmood, Saccomanno, & Hellinga, 2003) introduced the system dynamics method into a successful car following model, which takes into consideration the interactions of a following and two leading vehicles. Further, other techniques from artificial intelligence are being utilised in an effort to make car

Even though there were many efforts through the years to develop realistic models of car following behaviour there are significant limitations concerning their validity. Limitations arise mainly from unrealistic assumptions about the ability of drivers of following vehicles to perceive relative or absolute speeds and accelerations of the interacting vehicles. As Boer (Boer, 1999) suggests factors such as aging impairment and disability further influence driver reactions, which current car following models do not take into consideration. Further, unrealistic is the assumption that driver behaviour is influenced only by the immediate leading vehicle motion as observed by a number of researchers such as Fox and Lehman, and Bexelius. Also the assumption for an empirical relationship fails to explain actual behaviour as pointed out by (Van Winsum, 1999) and (Gipps, 1981) . Finally, existing car following models are rather idealistic as they assume symmetrical driver responses to traffic

stimuli, which is clearly unrealistic as revealed by (Chakroborty & Kikuchi, 1999).

models is given by (Brackstone & McDonald, 1999).

much less attention.

As (Brackstone & McDonald, 1999) conclude in their review on microscopic car following models there are potential pitfalls awaiting the unwary in the use of microcopic models. A comprehensive review on the weaknesses and potential developments of microscopic

Research on the existing models of driver behaviour has been restricted to modelling driver behaviour under car-following situations. Little work was found, on models of driver behaviour under various other driving situations. It can be said that most research so far has been concentrated on modelling driver behaviour in situations, where only longitudinal interactions affect the driver. Situations where, either lateral interactions alone or lateral as well as longitudinal interactions affect driver's behaviour received

separation distance of vehicle in front of the immediate leading vehicle.

leading vehicles as follows.

following models more realistic.

$$a\_f = F(v\_{l'}, v\_{f'}, \mathbf{s}\_{\prime} d\_{l'}, d\_f R\_f P\_i),$$

where *<sup>f</sup> a* is the acceleration (response) of the following vehicle, *<sup>l</sup> v* is the velocity of the leading vehicle, *<sup>f</sup> v* is the velocity of the following vehicle, *s* is the spacing between follower and leader vehicles or separation distance, *<sup>l</sup> d* and *<sup>f</sup> d* are the projected deceleration rates of the leader and follower vehicles, respectively, *Rf* is the reaction time of the driver in the following vehicle and *Pi* are other parameters specific to the car-following model. Based on this generalised equation several car following models have been developed through the years and are briefly presented below using the above general model notation.

(Chandler, Herman, & Montroll, 1958) developed a car following model assuming that the following vehicle driver responds solely to changes in the speed of the immediate leader vehicle. Chandler's model is given by the following equation.

$$a\_f = P\_1(v\_l - v\_f)$$

(Gazis, Herman, & Rothery, 1961) developed a more complex model adding the position of the leading and following vehicles in the equation and thereby introducing the notion of a safety distance between the two vehicles.

$$a\_f = \frac{P\_1}{d\_l - d\_f}(v\_l - v\_f)$$

(Edie, 1961) developed a similar response-stimulus model involving velocity and position changes as shown below.

$$a\_f = P\_1 \frac{V\_f}{\left(d\_l - d\_f\right)^2} (V\_l - V\_f)^2$$

(Herman & Rothery, 1962) apart from velocity and position changes included some other parameters in their car following model.

$$a\_f = P\_1 \frac{V\_f^{p\_2}}{\left(d\_l - d\_f\right)^{P\_3}} (V\_l - V\_f)$$

(Bierley, 1963) in a similar fashion but increasing the number of parameters suggested the following model.

$$a\_f = P\_1(\upsilon\_l - \upsilon\_f) + P\_2(d\_l - d\_f)$$

(Fox & Lehman, 1967) added another lead vehicle in their car following model by considering the changes in speed and position of the vehicle in front of the immediate leading vehicle as shown below.

$$a\_f = P\_1 V\_f \left(\frac{P\_2 \left(v\_{ll} - v\_f\right)}{\left(d\_{ll} - d\_f\right)^2} + \frac{P\_3 \left(v\_l - v\_f\right)}{\left(d\_l - d\_f\right)^2}\right)$$

( , ,, , ) *<sup>f</sup> <sup>l</sup> <sup>f</sup> <sup>l</sup> f f <sup>i</sup> a Fv v sd dRP*

where *<sup>f</sup> a* is the acceleration (response) of the following vehicle, *<sup>l</sup> v* is the velocity of the leading vehicle, *<sup>f</sup> v* is the velocity of the following vehicle, *s* is the spacing between follower and leader vehicles or separation distance, *<sup>l</sup> d* and *<sup>f</sup> d* are the projected deceleration rates of the leader and follower vehicles, respectively, *Rf* is the reaction time of the driver in the following vehicle and *Pi* are other parameters specific to the car-following model. Based on this generalised equation several car following models have been developed through the years

(Chandler, Herman, & Montroll, 1958) developed a car following model assuming that the following vehicle driver responds solely to changes in the speed of the immediate leader

1( ) *<sup>f</sup> <sup>l</sup> <sup>f</sup> a Pv v*

(Gazis, Herman, & Rothery, 1961) developed a more complex model adding the position of the leading and following vehicles in the equation and thereby introducing the notion of a

> <sup>1</sup> ( ) *<sup>f</sup> <sup>l</sup> <sup>f</sup> l f P a vv d d*

(Edie, 1961) developed a similar response-stimulus model involving velocity and position

(Herman & Rothery, 1962) apart from velocity and position changes included some other

2 <sup>3</sup> <sup>1</sup> ( ) ( )

(Bierley, 1963) in a similar fashion but increasing the number of parameters suggested the

1 2 ( )( ) *<sup>f</sup> <sup>l</sup> <sup>f</sup> <sup>l</sup> <sup>f</sup> a Pv v Pd d*

(Fox & Lehman, 1967) added another lead vehicle in their car following model by considering the changes in speed and position of the vehicle in front of the immediate

> 2 3 1 2 2 ( )( ) ( ) ( )( ) *ll f l f*

*ll f l f Pv v Pv v*

*d d dd* 

*f f*

*a PV*

*P f f P l f l f*

*V a P VV d d*

<sup>1</sup> <sup>2</sup> ( ) ( ) *f f l f l f V a P VV d d*

and are briefly presented below using the above general model notation.

vehicle. Chandler's model is given by the following equation.

safety distance between the two vehicles.

parameters in their car following model.

changes as shown below.

following model.

leading vehicle as shown below.

where *ll v* the speed of the vehicle in front of the immediate leading vehicle and *ll d* the separation distance of vehicle in front of the immediate leading vehicle.

Bexelius (Bexelius, 1968) also suggested a car following model that takes into account two leading vehicles as follows.

$$a\_f = P\_1(v\_{ll} - v\_f) + P\_2(v\_l - v\_f)$$

More complex car following models of similar nature were developed by (Wicks, Lieberman, Associates, & Division, 1980) where the NETSIM software is based, and by (Bullen, 1982) where FRESIM is based. Further, (Gipps, 1981) developed a safety distance or behavioural model, which is employed by AIMSUN. (Fritzsche, 1994) and (Wiedemann, 1974) developed the so called action point or phycho-physical models for Paramics and VISSIM respectively. Some more modern approaches to car following models make use of fuzzy logic algorithms (Gonzalez-Rojo, Slama, Lopes, & Mora-Camino, 2002; Yikai, Satoh, Itakura, Honda, & Satoh, 1993). Interesting is also the employment of the System Dynamics modelling principles in car following. More recently, (Mehmood, Saccomanno, & Hellinga, 2003) introduced the system dynamics method into a successful car following model, which takes into consideration the interactions of a following and two leading vehicles. Further, other techniques from artificial intelligence are being utilised in an effort to make car following models more realistic.

Even though there were many efforts through the years to develop realistic models of car following behaviour there are significant limitations concerning their validity. Limitations arise mainly from unrealistic assumptions about the ability of drivers of following vehicles to perceive relative or absolute speeds and accelerations of the interacting vehicles. As Boer (Boer, 1999) suggests factors such as aging impairment and disability further influence driver reactions, which current car following models do not take into consideration. Further, unrealistic is the assumption that driver behaviour is influenced only by the immediate leading vehicle motion as observed by a number of researchers such as Fox and Lehman, and Bexelius. Also the assumption for an empirical relationship fails to explain actual behaviour as pointed out by (Van Winsum, 1999) and (Gipps, 1981) . Finally, existing car following models are rather idealistic as they assume symmetrical driver responses to traffic stimuli, which is clearly unrealistic as revealed by (Chakroborty & Kikuchi, 1999).

As (Brackstone & McDonald, 1999) conclude in their review on microscopic car following models there are potential pitfalls awaiting the unwary in the use of microcopic models. A comprehensive review on the weaknesses and potential developments of microscopic models is given by (Brackstone & McDonald, 1999).

Research on the existing models of driver behaviour has been restricted to modelling driver behaviour under car-following situations. Little work was found, on models of driver behaviour under various other driving situations. It can be said that most research so far has been concentrated on modelling driver behaviour in situations, where only longitudinal interactions affect the driver. Situations where, either lateral interactions alone or lateral as well as longitudinal interactions affect driver's behaviour received much less attention.

Modelling, Simulation Methods for Intelligent Transportation Systems 113

In today's traffic simulation software, data such as network definition of roads and tracks, technical vehicle and behavioural driver specifications, car volumes and paths can be inserted in graphical user interface mode. Values for acceleration, maximum speed and desired speed distributions can be configured by the user to reflect local traffic conditions. Various vehicles types can also be defined. Further, traffic control strategies and algorithms may be defined as well as interfaces may be built with well-known urban traffic controllers. CORSIM, PARAMICS, VISSIM, and AIMSUN were calibrated and validated in a number of traffic studies worldwide. Below we present some of their main

CORSIM which stands for Corridor microscopic simulation is developed by Federal Highway Administration of United States. It has evolved from two separate traffic simulation programs NETSIM and FRESIM. NETSIM models arterials with signalised and unsignalised intersections, while FRESIM models uninterrupted freeways and urban

In the case of VISSIM the microscopic model consists of a psycho-physical car following model for longitudinal vehicle movement and a rule-based lane changing algorithm for lateral movements. The model is based on an urban and a freeway model which were developed by Wiedemann from the University of Karlsruhe. VISSIM is especially well known for its signal control module, which uses a vehicle actuated programming language can model almost any traffic control logic. Further, VISSIM scores high on its ability to

AIMSUN was developed by TSS in order to simulate urban and interurban traffic networks. It is based on the car-following model of Gibbs . AIMSUN is therefore based on a collision avoidance car-following model. Traffic can be modelled via input flows and turning

PARAMICS, which stands for Parallel Microscopic Simulation, comprises of various modules which include a modeller, a processor, an analyser, a monitor, a converter and an estimator. PARAMICS is well known for its visualization graphics and for its ability to

A comprehensive review of simulation models of traffic flow was conducted by the Institute for Transport Studies at the University of Leeds as part of the SMARTEST Project which is a collaborative project to develop micro-simulation tools to help solve road traffic management problems. The study compared the capabilities of more than 50 simulation packages. The results are available on the internet at http://www.its. leeds.ac.uk/ projects/smartest. Other significant reviews of traffic simulation software include the work of (Bloomberg & Dale, 2000) who compared Corsim and Vissim as well as the work of (Boxill, Yu, Training, Research, & Center, 2000) who compared the capabilities of Corsim, Aimsun and Paramics. It can be concluded from the various reviews that software modelers that have comparative capabilities include VISSIM,

In a more recent comparative study of microscopic car following behavior, (Panwai & Dia, 2005) evaluate AIMSUN, VISSIM and PARAMICS. They concluded that the accuracy of a

movements, origin destination matrices, and route choice models.

features.

highways.

model public transportation systems.

model quite a diverse range of traffic scenarios.

AIMSUN, and PARAMICS.

## **3. Microscopic traffic modelling software tools**

Microscopic simulation is a term used in traffic modelling and is typified by software packages such as VISSIM (Fellendorf & Vortisch, 2001; Gomes, May, & Horowitz, 2004; Park, Won, & Yun, 2006; PTV, 2005), CORSIM (Lin, 1998; Prevedouros & Wang, 1999; Zhang, McHale, & Zhang, 2003), and PARAMICS (Gardes, 2006; Jacob & Abdulhai, 2006; Ozbay, Bartin, Mudigonda, & Board, 2006). Traffic simulation microscopic models simulate the behaviour of individual vehicles within a predefined road network and are used to predict the likely impact of changes in traffic patterns resulting from proposed commercial developments or road schemes. They are aiming to facilitate transportation consultants, municipalities, government transportation authorities and public transportation companies. The traffic flow models used are discrete, stochastic, time step based microscopic models, with driver-vehicle units as single entities.

Traffic simulation software modelers combine in a single package multiple traffic flow mathematical models and therefore make it possible to combine the current knowledge on traffic theory when analyzing a traffic congestion problem. A screenshot of the VISSIM graphical user interface is provided in figure 2. The microscopic model depicted in the figure was developed in order to analyze traffic and evaluate the impact of various bus priority scenarios for a traffic network in Nicosia, Cyprus (G. Papageorgiou, 2006; G. Papageorgiou, Damianou, Pitsillides, Aphames, & Ioannou, 2006).

Fig. 2. VISSIM Graphical User Interface depicting part of the microscopic model of Strovolos Ave. in Nicosia, Cyprus.

Microscopic simulation is a term used in traffic modelling and is typified by software packages such as VISSIM (Fellendorf & Vortisch, 2001; Gomes, May, & Horowitz, 2004; Park, Won, & Yun, 2006; PTV, 2005), CORSIM (Lin, 1998; Prevedouros & Wang, 1999; Zhang, McHale, & Zhang, 2003), and PARAMICS (Gardes, 2006; Jacob & Abdulhai, 2006; Ozbay, Bartin, Mudigonda, & Board, 2006). Traffic simulation microscopic models simulate the behaviour of individual vehicles within a predefined road network and are used to predict the likely impact of changes in traffic patterns resulting from proposed commercial developments or road schemes. They are aiming to facilitate transportation consultants, municipalities, government transportation authorities and public transportation companies. The traffic flow models used are discrete, stochastic, time step

Traffic simulation software modelers combine in a single package multiple traffic flow mathematical models and therefore make it possible to combine the current knowledge on traffic theory when analyzing a traffic congestion problem. A screenshot of the VISSIM graphical user interface is provided in figure 2. The microscopic model depicted in the figure was developed in order to analyze traffic and evaluate the impact of various bus priority scenarios for a traffic network in Nicosia, Cyprus (G. Papageorgiou, 2006; G.

Fig. 2. VISSIM Graphical User Interface depicting part of the microscopic model of Strovolos

Ave. in Nicosia, Cyprus.

**3. Microscopic traffic modelling software tools** 

based microscopic models, with driver-vehicle units as single entities.

Papageorgiou, Damianou, Pitsillides, Aphames, & Ioannou, 2006).

In today's traffic simulation software, data such as network definition of roads and tracks, technical vehicle and behavioural driver specifications, car volumes and paths can be inserted in graphical user interface mode. Values for acceleration, maximum speed and desired speed distributions can be configured by the user to reflect local traffic conditions. Various vehicles types can also be defined. Further, traffic control strategies and algorithms may be defined as well as interfaces may be built with well-known urban traffic controllers. CORSIM, PARAMICS, VISSIM, and AIMSUN were calibrated and validated in a number of traffic studies worldwide. Below we present some of their main features.

CORSIM which stands for Corridor microscopic simulation is developed by Federal Highway Administration of United States. It has evolved from two separate traffic simulation programs NETSIM and FRESIM. NETSIM models arterials with signalised and unsignalised intersections, while FRESIM models uninterrupted freeways and urban highways.

In the case of VISSIM the microscopic model consists of a psycho-physical car following model for longitudinal vehicle movement and a rule-based lane changing algorithm for lateral movements. The model is based on an urban and a freeway model which were developed by Wiedemann from the University of Karlsruhe. VISSIM is especially well known for its signal control module, which uses a vehicle actuated programming language can model almost any traffic control logic. Further, VISSIM scores high on its ability to model public transportation systems.

AIMSUN was developed by TSS in order to simulate urban and interurban traffic networks. It is based on the car-following model of Gibbs . AIMSUN is therefore based on a collision avoidance car-following model. Traffic can be modelled via input flows and turning movements, origin destination matrices, and route choice models.

PARAMICS, which stands for Parallel Microscopic Simulation, comprises of various modules which include a modeller, a processor, an analyser, a monitor, a converter and an estimator. PARAMICS is well known for its visualization graphics and for its ability to model quite a diverse range of traffic scenarios.

A comprehensive review of simulation models of traffic flow was conducted by the Institute for Transport Studies at the University of Leeds as part of the SMARTEST Project which is a collaborative project to develop micro-simulation tools to help solve road traffic management problems. The study compared the capabilities of more than 50 simulation packages. The results are available on the internet at http://www.its. leeds.ac.uk/ projects/smartest. Other significant reviews of traffic simulation software include the work of (Bloomberg & Dale, 2000) who compared Corsim and Vissim as well as the work of (Boxill, Yu, Training, Research, & Center, 2000) who compared the capabilities of Corsim, Aimsun and Paramics. It can be concluded from the various reviews that software modelers that have comparative capabilities include VISSIM, AIMSUN, and PARAMICS.

In a more recent comparative study of microscopic car following behavior, (Panwai & Dia, 2005) evaluate AIMSUN, VISSIM and PARAMICS. They concluded that the accuracy of a

Modelling, Simulation Methods for Intelligent Transportation Systems 115

Fig. 3. The Archangelou Avenue Case Study Area with the nearby Traffic Network.

Research Ltd.

The aim of the study titled "Intelligent Transportation Systems in Archangelou Avenue (BUSSIM)" was to develop and test BRT strategies via scenario analysis in a computer simulated environment. Scenarios that were evaluated include a number of configurations regarding the introduction of dedicated bus lanes as well as bus advance signal areas as well as High Occupancy Vehicle (HOV) lanes. The scenario analysis was carried out via computer experiments using a microscopic simulation model of Archangelou Avenue urban traffic network. The case study presented in this chapter is part of the BUSSIM research project (George Papageorgiou, Maimaris, Ioannou, Pitsillides, & Afamis, 2010) which was funded by the Cyprus Research Promotion Foundation and Transim Transportation

**Study Area** 

As shown in figure 4, the first step of the proposed approach is to identify and define the problem. In our case the symptoms of the problem which are attributed to traffic congestion manifest themselves as increasing travel times for all transport modes. The main causes to the problem of traffic congestion in Nicosia consist of an increasing number of vehicles and a decreasing use of the bus transportation system. Adding more capacity to the road infrastructure will only make things worse, as a reinforcing feedback loop is created where further use of private vehicles is encouraged and use of the public transport is discouraged. Therefore, the long term solution to the problem is to balance or even to turn around the

situation by encouraging the use of the public transport mode.

traffic simulation system depends highly on the quality of its traffic flow model at its core, which consists of car following and lane changing models. In the study the car-following behaviour for each simulator was compared to field data obtained from instrumented vehicles travelling on an urban road in Germany. The Error Metric on distance (Manstetten, Krautter, & Schwab, 1997) performance indicator gave substantially better values for AIMSUN than those of VISSIM and PARAMICS. Further, the Root Mean Square Error (RMSE) was substantially less for VISSIM and AIMSUN than the RMSE for PARAMICS. In another paper presented at the 9th TRB Conference on the Application of Transportation Planning Methods Choa et al. (Choa, Milam, & Stanek, 2004) concluded that although CORSIM provides the shortest traffic network setup time , PARAMICS and VISSIM generated simulation results that better matched field observed conditions and traffic engineering principles.

The reason microscopic simulation models are used over other software packages and methods like the Highway Capacity Manual (HCM) is that microscopic simulations allow us to evaluate the effects that different traffic elements have on each other. Being able to evaluate the effects of closely spaced intersections and interchanges on the traffic network or the effects of a bottleneck condition on the surrounding system, can only be achieved by microscopic traffic simulation models. Also, as metropolitan traffic conditions experience congestion over 3 to 4 hour periods, microscopic traffic simulation programs allow us to evaluate the build-up to congested conditions and the recovery of the system at the end of the period. The peak period of congestion is complex and evaluating solutions under these conditions can only be accomplished using microscopic simulation tools.

In the following section, an approach to modelling and simulation of intelligent transportation systems (ITS) is proposed and implemented for a particular case study in Nicosia , Cyprus. The approach utilizes the VISSIM microscopic simulation modeler.

## **4. ITS studies using microscopic simulation**

As described in the previous sections traffic phenomena constitute a dynamical problem situation, which makes traffic modeling and simulation a very complex, iterative and tedious process. In order to increase chances for developing a realistic simulation model the following methodology is developed, which is based on the suggestions of (Lieberman & Rathi, 1996) and (Dowling, 2007). This is applied for the modeling of Archangelou Avenue traffic network in Nicosia, Cyprus as described below.

The study area is depicted in figure 3, which shows Archangelou Avenue with its nearby traffic network. Archangelou Avenue is the main road connecting the Rural Nicosia District to the centre of Nicosia, where the main business center is located. Nicosia, the capital of Cyprus, has a population of around 350,000 people. Archangelou Ave., which is one of the three main arterial roads exhibits very high traffic flows as compared with the other regions of metropolitan Nicosia. Further, Archangelou Avenue serves as the connector between Nicosia and a large and heavily populated area of urban and rural communities.

traffic simulation system depends highly on the quality of its traffic flow model at its core, which consists of car following and lane changing models. In the study the car-following behaviour for each simulator was compared to field data obtained from instrumented vehicles travelling on an urban road in Germany. The Error Metric on distance (Manstetten, Krautter, & Schwab, 1997) performance indicator gave substantially better values for AIMSUN than those of VISSIM and PARAMICS. Further, the Root Mean Square Error (RMSE) was substantially less for VISSIM and AIMSUN than the RMSE for PARAMICS. In another paper presented at the 9th TRB Conference on the Application of Transportation Planning Methods Choa et al. (Choa, Milam, & Stanek, 2004) concluded that although CORSIM provides the shortest traffic network setup time , PARAMICS and VISSIM generated simulation results that better matched field observed conditions and traffic

The reason microscopic simulation models are used over other software packages and methods like the Highway Capacity Manual (HCM) is that microscopic simulations allow us to evaluate the effects that different traffic elements have on each other. Being able to evaluate the effects of closely spaced intersections and interchanges on the traffic network or the effects of a bottleneck condition on the surrounding system, can only be achieved by microscopic traffic simulation models. Also, as metropolitan traffic conditions experience congestion over 3 to 4 hour periods, microscopic traffic simulation programs allow us to evaluate the build-up to congested conditions and the recovery of the system at the end of the period. The peak period of congestion is complex and evaluating solutions under these conditions can only be accomplished using microscopic simulation

In the following section, an approach to modelling and simulation of intelligent transportation systems (ITS) is proposed and implemented for a particular case study in Nicosia , Cyprus. The approach utilizes the VISSIM microscopic simulation modeler.

As described in the previous sections traffic phenomena constitute a dynamical problem situation, which makes traffic modeling and simulation a very complex, iterative and tedious process. In order to increase chances for developing a realistic simulation model the following methodology is developed, which is based on the suggestions of (Lieberman & Rathi, 1996) and (Dowling, 2007). This is applied for the modeling of Archangelou Avenue

The study area is depicted in figure 3, which shows Archangelou Avenue with its nearby traffic network. Archangelou Avenue is the main road connecting the Rural Nicosia District to the centre of Nicosia, where the main business center is located. Nicosia, the capital of Cyprus, has a population of around 350,000 people. Archangelou Ave., which is one of the three main arterial roads exhibits very high traffic flows as compared with the other regions of metropolitan Nicosia. Further, Archangelou Avenue serves as the connector between Nicosia and a large and heavily populated area of urban and rural

**4. ITS studies using microscopic simulation** 

traffic network in Nicosia, Cyprus as described below.

engineering principles.

tools.

communities.

Fig. 3. The Archangelou Avenue Case Study Area with the nearby Traffic Network.

The aim of the study titled "Intelligent Transportation Systems in Archangelou Avenue (BUSSIM)" was to develop and test BRT strategies via scenario analysis in a computer simulated environment. Scenarios that were evaluated include a number of configurations regarding the introduction of dedicated bus lanes as well as bus advance signal areas as well as High Occupancy Vehicle (HOV) lanes. The scenario analysis was carried out via computer experiments using a microscopic simulation model of Archangelou Avenue urban traffic network. The case study presented in this chapter is part of the BUSSIM research project (George Papageorgiou, Maimaris, Ioannou, Pitsillides, & Afamis, 2010) which was funded by the Cyprus Research Promotion Foundation and Transim Transportation Research Ltd.

As shown in figure 4, the first step of the proposed approach is to identify and define the problem. In our case the symptoms of the problem which are attributed to traffic congestion manifest themselves as increasing travel times for all transport modes. The main causes to the problem of traffic congestion in Nicosia consist of an increasing number of vehicles and a decreasing use of the bus transportation system. Adding more capacity to the road infrastructure will only make things worse, as a reinforcing feedback loop is created where further use of private vehicles is encouraged and use of the public transport is discouraged. Therefore, the long term solution to the problem is to balance or even to turn around the situation by encouraging the use of the public transport mode.

Modelling, Simulation Methods for Intelligent Transportation Systems 117

DBL 4

BA 3

BA 4

DBL 3

Fig. 5. The Traffic Simulation Model Showing The Proposed Road Design for Dedicated Bus

BA 2

The model incorporated a significant amount of various traffic data that may be classified in terms of static data and dynamic data. Static data represents the roadway infrastructure. It includes links, which are directional roadway segments with a specified number of lanes, with start and end points as well as optional intermediate points. Further, static data includes connectors between links, which are used to model turnings, lane drops and lane gains, locations and length of transit stops, position of signal heads/stop lines including a reference to the associated signal group, and positions and length of detectors. Dynamic data was also specified for the traffic simulation experiments. It included traffic volumes with vehicle mix for all links entering the network, locations of route decision points with routes, that is the link sequences to be followed, differentiated by time and vehicle classification, priority rules, right-of-way to model un-signalized intersections, permissive turns at signalized junctions and yellow boxes or keep-clear-areas, locations of stop signs,

Having introduced the necessary traffic parameters in the model, the iterative process begun, which consisted of model development calibration and validation of the model.

Figure 6 shows the real Vs simulated traffic flows of the various vehicle movement directions of a central intersection of Archangelou Avenue, in particular that of Archangelou-Odyssea Elyti. As seen in the bar chart, traffic flows of real measurements obtained and those of simulated results, are quite comparable. In particular the error ranges from only 1% to 5%, a fact that contributes to building confidence for the model. Further, the simulation model demonstrated the queues that are encountered in reality during the

Lane (DBL) and Bus Advance Areas (BA).

DBL 2

BA 1

DBL 1

morning peak hours.

public transport routing, departure times and dwell times.

Fig. 4. The Proposed Traffic Modeling and Simulation Method (in paper by Papageorgiou et al presented at 12th IFAC Symposium on Transportation Systems, September 2009).

The question then becomes how to attract people in using the bus transportation system. The answer to this was given by the citizens of Cyprus in a recent survey where they expressed their wish for higher quality, faster public bus transport system. This is what was investigated in the BUSSIM project, concentrating on providing a faster and better quality level of service for bus passengers. The objective therefore in the modeling and simulation method was to examine various scenarios such as dedicated bus lanes and Bus Rapid Transit Systems that would provide a better level of service for the bus transportation system. Meanwhile there was a need to anticipate and assess any side effects of plausible solutions to the rest of the transportation system.

Based on the stated model objectives, a microscopic simulation model of Archangelou Avenue was developed. Like any other traffic network, Archangelou Avenue consisted of many traffic parameters that needed to be taken into account. These included traffic control signals, priority rules, routing decisions, and pedestrian crossings, signalized and unsignalized intersections and so on. A helicopter view of the simulation model of Archangelou Avenue is depicted in figure 5 (see also figure 3). Figure 5 shows the proposed layout of Archangelou Ave, which is a five-lane road more than 4 kilometers long, as well as the main roads that intersect Archangelou Avenue. Figure 5 also shows potential areas for introducing dedicated bus lanes.

Fig. 4. The Proposed Traffic Modeling and Simulation Method (in paper by Papageorgiou et

The question then becomes how to attract people in using the bus transportation system. The answer to this was given by the citizens of Cyprus in a recent survey where they expressed their wish for higher quality, faster public bus transport system. This is what was investigated in the BUSSIM project, concentrating on providing a faster and better quality level of service for bus passengers. The objective therefore in the modeling and simulation method was to examine various scenarios such as dedicated bus lanes and Bus Rapid Transit Systems that would provide a better level of service for the bus transportation system. Meanwhile there was a need to anticipate and assess any side effects of plausible

Based on the stated model objectives, a microscopic simulation model of Archangelou Avenue was developed. Like any other traffic network, Archangelou Avenue consisted of many traffic parameters that needed to be taken into account. These included traffic control signals, priority rules, routing decisions, and pedestrian crossings, signalized and unsignalized intersections and so on. A helicopter view of the simulation model of Archangelou Avenue is depicted in figure 5 (see also figure 3). Figure 5 shows the proposed layout of Archangelou Ave, which is a five-lane road more than 4 kilometers long, as well as the main roads that intersect Archangelou Avenue. Figure 5 also shows potential areas for

al presented at 12th IFAC Symposium on Transportation Systems, September 2009).

solutions to the rest of the transportation system.

introducing dedicated bus lanes.

Fig. 5. The Traffic Simulation Model Showing The Proposed Road Design for Dedicated Bus Lane (DBL) and Bus Advance Areas (BA).

The model incorporated a significant amount of various traffic data that may be classified in terms of static data and dynamic data. Static data represents the roadway infrastructure. It includes links, which are directional roadway segments with a specified number of lanes, with start and end points as well as optional intermediate points. Further, static data includes connectors between links, which are used to model turnings, lane drops and lane gains, locations and length of transit stops, position of signal heads/stop lines including a reference to the associated signal group, and positions and length of detectors. Dynamic data was also specified for the traffic simulation experiments. It included traffic volumes with vehicle mix for all links entering the network, locations of route decision points with routes, that is the link sequences to be followed, differentiated by time and vehicle classification, priority rules, right-of-way to model un-signalized intersections, permissive turns at signalized junctions and yellow boxes or keep-clear-areas, locations of stop signs, public transport routing, departure times and dwell times.

Having introduced the necessary traffic parameters in the model, the iterative process begun, which consisted of model development calibration and validation of the model.

Figure 6 shows the real Vs simulated traffic flows of the various vehicle movement directions of a central intersection of Archangelou Avenue, in particular that of Archangelou-Odyssea Elyti. As seen in the bar chart, traffic flows of real measurements obtained and those of simulated results, are quite comparable. In particular the error ranges from only 1% to 5%, a fact that contributes to building confidence for the model. Further, the simulation model demonstrated the queues that are encountered in reality during the morning peak hours.

Modelling, Simulation Methods for Intelligent Transportation Systems 119

This chapter presented an overview of the most important developments in traffic flow theory, and examines modelling of traffic flow at two fundamental levels: the macroscopic level, where traffic is regarded as a fluid, and the microscopic level, where traffic is represented by individual driver-vehicle units. Concerning these two levels of analysis, without discarding the usefulness of macroscopic models it may be concluded that as a result of advancements in computer technology, and the need for more detailed and accurate traffic models there is a trend nowadays for microscopic traffic models where the ultimate goal is to capture the driver-vehicle unit interactions under a variety of driving

Further, this chapter provided an insight analysis to the world's most sophisticated traffic simulation modeller software, VISSIM, AIMSUN, CORSIM and PARAMICS, where their capabilities and limitations are discussed. Also, an approach to modeling and simulation of intelligent transportation systems is proposed and implemented. The proposed approach goes through various stages, which include problem identification, model objectives, model development, model calibration, model validation, scenario preparation, simulation experiments and simulated results evaluation. The proposed approach is applied in the case of developing a microscopic traffic simulation model for the urban traffic network of Archangelou Avenue, Nicosia, Cyprus in order to examine alternative bus transport mode

Adams, W. F. (1937). *Road traffic considered as a random series*: Institution of Civil Engineers. Ardekani, S., Hauer, E., & Jamei, B. (1992). Traffic impact models. *Chapter 7 in Traffic Flow* 

Bexelius, S. (1968). An extended model for car-following. *Transportation Research, 2*(1), 13-21. Bierley, R. L. (1963). Investigation of an Intervehicle Spacing Display. *Highway Research* 

Bloomberg, L., & Dale, J. (2000). Comparison of VISSIM and CORSIM traffic simulation

Boer, E. R. (1999). Car following from the driver's perspective. *Transportation Research Part F:* 

Boxill, S. A., Yu, L., Training, T. S. U. C. f. T., Research, & Center, S. R. U. T. (2000). *An* 

Brackstone, M., & McDonald, M. (1999). Car-following: a historical review. *Transportation* 

Bullen, A. (1982). Development of Compact Microsimulation for Analyzing Freeway

Chakroborty, P., & Kikuchi, S. (1999). Evaluation of the General Motors based car-following

Chandler, R. E., Herman, R., & Montroll, E. W. (1958). Traffic dynamics: studies in car

Transportation Training and Research, Texas Southern University.

*Research Part F: Traffic Psychology and Behaviour, 2*(4), 181-196.

Operations and Design. *Transportation Research Record*(841).

models on a congested network. *Transportation Research Record: Journal of the* 

*evaluation of traffic simulation models for supporting ITS development*: Center for

models and a proposed fuzzy inference model. *Transportation Research Part C:* 

conditions in a computer simulated environment.

**6. References** 

*Record*.

enhancements by means of Intelligent Transportation Systems.

*Theory, Oak Bridge National Laboratory Report*.

*Transportation Research Board, 1727*(-1), 52-60.

*Traffic Psychology and Behaviour, 2*(4), 201-206. Bose, A., & Ioannou, P. (2000). *Shock Waves in Mixed Traffic Flow*.

*Emerging Technologies, 7*(4), 209-235.

following. *Operations Research*, 165-184.

Fig. 6. Model Validation: Simulated Vs Measured flows at Archangelou-Odyssea Elyti junction.

With a validated model in our hands, next comes the preparation of BRT scenarios, their evaluation and the analysis of the results. After consultations with the transportation planning section of the Ministry of Communications Works we came up with several plausible scenarios. In summary, the various scenarios involve the use of dedicated bus lanes and High Occupancy Vehicle lanes by means of Intelligent Transportation Systems.

Even though the modelling process and especially the calibration of the microscopic model was time consuming, results from the simulation experiments gave us significant information on a variety of measures of effectiveness (MoE). In particular, we have managed to compute travel times, queue lengths, delays, average speeds, lane changes and other MoEs for the various scenarios under investigation. On the basis of the various MoEs comparison was carried out between the various scenarios using hypothesis testing with a 95% confidence interval (results submitted for publication). Such valuable information is obviously essential for the implementation of any Intelligent Transportation Systems project.

#### **5. Conclusion**

Modelling and simulation methods are essential elements in the design, operation and control of Intelligent Transportation Systems (ITS). Congestion problems in cities worldwide have drawn a high level of interest for better management and control of transportation systems. Of major importance are ITS systems that include advanced traffic management and control techniques. Such techniques include real-time traffic control measures and realtime traveller information and guidance systems whose purpose is to assist travellers in making departure time, mode and route choice decisions. Transportation research is heading towards developing models and simulators for use in the planning, design and operations and control of such intelligent transportation systems.

This chapter presented an overview of the most important developments in traffic flow theory, and examines modelling of traffic flow at two fundamental levels: the macroscopic level, where traffic is regarded as a fluid, and the microscopic level, where traffic is represented by individual driver-vehicle units. Concerning these two levels of analysis, without discarding the usefulness of macroscopic models it may be concluded that as a result of advancements in computer technology, and the need for more detailed and accurate traffic models there is a trend nowadays for microscopic traffic models where the ultimate goal is to capture the driver-vehicle unit interactions under a variety of driving conditions in a computer simulated environment.

Further, this chapter provided an insight analysis to the world's most sophisticated traffic simulation modeller software, VISSIM, AIMSUN, CORSIM and PARAMICS, where their capabilities and limitations are discussed. Also, an approach to modeling and simulation of intelligent transportation systems is proposed and implemented. The proposed approach goes through various stages, which include problem identification, model objectives, model development, model calibration, model validation, scenario preparation, simulation experiments and simulated results evaluation. The proposed approach is applied in the case of developing a microscopic traffic simulation model for the urban traffic network of Archangelou Avenue, Nicosia, Cyprus in order to examine alternative bus transport mode enhancements by means of Intelligent Transportation Systems.

#### **6. References**

118 Intelligent Transportation Systems

Fig. 6. Model Validation: Simulated Vs Measured flows at Archangelou-Odyssea Elyti

With a validated model in our hands, next comes the preparation of BRT scenarios, their evaluation and the analysis of the results. After consultations with the transportation planning section of the Ministry of Communications Works we came up with several plausible scenarios. In summary, the various scenarios involve the use of dedicated bus lanes and High Occupancy Vehicle lanes by means of Intelligent Transportation Systems.

Even though the modelling process and especially the calibration of the microscopic model was time consuming, results from the simulation experiments gave us significant information on a variety of measures of effectiveness (MoE). In particular, we have managed to compute travel times, queue lengths, delays, average speeds, lane changes and other MoEs for the various scenarios under investigation. On the basis of the various MoEs comparison was carried out between the various scenarios using hypothesis testing with a 95% confidence interval (results submitted for publication). Such valuable information is obviously essential

Modelling and simulation methods are essential elements in the design, operation and control of Intelligent Transportation Systems (ITS). Congestion problems in cities worldwide have drawn a high level of interest for better management and control of transportation systems. Of major importance are ITS systems that include advanced traffic management and control techniques. Such techniques include real-time traffic control measures and realtime traveller information and guidance systems whose purpose is to assist travellers in making departure time, mode and route choice decisions. Transportation research is heading towards developing models and simulators for use in the planning, design and

for the implementation of any Intelligent Transportation Systems project.

operations and control of such intelligent transportation systems.

junction.

**5. Conclusion** 

Adams, W. F. (1937). *Road traffic considered as a random series*: Institution of Civil Engineers.


Modelling, Simulation Methods for Intelligent Transportation Systems 121

Kikuchi, S., Chakroborty, P., & Engineering, U. o. D. D. o. C. (1992). *A car-following model* 

Koppa, R. J. (1999). Human factors. *Traffic Flow Theory, a state of the art report. Revised Monograph on Traffic Flow Theory", ed. by: Gartner, N., CJ Messer & AK Rathi*. Kuhne, R., & Michalopoulos, P. (1997). Continuum flow models. *Traffic flow theory: A state of* 

Lieberman, E., & Rathi, A. (1996). Traffic Simulation. in Traffic Flow Theory. *Washington,* 

Lighthill, M. J., & Whitham, G. B. (1955). On kinematic waves. II. A theory of traffic flow on

Lin, S. (1998). CORSIM micro-node logic: Technical Report, Federal Highway

Manstetten, D., Krautter, W., & Schwab, T. (1997). *Traffic simulation supporting urban control* 

Mehmood, A., Saccomanno, F., & Hellinga, B. (2003). Application of system dynamics in car-

Michalopoulos, P. G., Yi, P., & Lyrintzis, A. S. (1992). Development of an improved highorder continuum traffic flow model. *Transportation Research Record*(1365). Nagel, K. (1996). Particle hopping models and traffic flow theory. *Physical Review E, 53*(5),

Nagel, K., & Schreckenberg, M. (1992). A cellular automaton model for freeway traffic.

Naiem, A., Reda, M., El-Beltagy, M., & El-Khodary, I. (2010) *An agent based approach for* 

Ozbay, K., Bartin, B. O., Mudigonda, S., & Board, T. R. (2006). *Microscopic Simulation and* 

Panwai, S., & Dia, H. (2005). Comparative evaluation of microscopic car-following behavior. *Intelligent Transportation Systems, IEEE Transactions on, 6*(3), 314-325. Papageorgiou, G. (2006). *Towards a microscopic simulation model for traffic management: a* 

Papageorgiou, G., Damianou, P., Pitsillides, A., Aphames, T., & Ioannou, P. (2006). *A Microscopic Traffic Simulation Model for Transportation Planning in Cyprus*. Papageorgiou, G., Maimaris, A., Ioannou, P., Pitsillides, A., & Afamis, T. (2010). *Introduction* 

Papageorgiou, M., Blosseville, J. M., & Hadj-Salem, H. (1989). Macroscopic modelling of

Park, B., Won, J., & Yun, I. (2006). Application of microscopic simulation model Calibration

*of Bus Rapid Transit in Cyprus: Evaluation of Bus Priority Scenarios.* Paper presented at

traffic flow on the Boulevard Périphérique in Paris. *Transportation Research Part B:* 

and validation procedure: Case study of coordinated actuated signal system. *Transportation Research Record: Journal of the Transportation Research Board, 1978*(-1),

following models. *Journal of transportation engineering, 129*, 625.

long crowded roads. *Proceedings of the Royal Society of London. Series A. Mathematical* 

*the art reportrevised monograph on traffic flow theory*.

*DC: US Federal Highway Administration*, 10-11.

*and Physical Sciences, 229*(1178), 317.

May, A. D. (1990). *Traffic flow fundamentals*: Prentice Hall.

*Journal de Physique I, 2*(12), 2221-2229.

*Calibration of Integrated Freeway and Toll Plaza Model*.

the Transportation Research Board 89th Annual Meeting.

Administration, McLean, VA.

*system development*.

*modeling traffic flow*.

*computer based approach*.

*Methodological, 23*(1), 29-47.

113-122.

4655.

Council.

*based on fuzzy inference system*: Transportation Research Board, National Research


Choa, F., Milam, R. T., & Stanek, D. (2004). *CORSIM, PARAMICS, and VISSIM: What the* 

Dowling, R. (2007). Traffic Analysis Toolbox Volume VI: Definition, Interpretation, and

Drake, J., Schofer, J., & May Jr, A. (1967). A statistical analysis of speed density hypotheses,

Edie, L. C. (1961). Car-following and steady-state theory for noncongested traffic. *Operations* 

Fellendorf, M., & Vortisch, P. (2001). *Validation of the microscopic traffic flow model VISSIM in* 

Forbes, T. (1963). Human factor considerations in traffic flow theory. *Highway Research* 

Fox, P., & Lehman, F. G. (1967). *Safety in car following: a computer simulation*: Newark College

Fritzsche, H. T. (1994). A model for traffic simulation. *Traffic Engineering and Control, 35*(5),

Gardes, Y. (2006). *Evaluating Traffic Calming and Capacity Improvements on SR-20 Corridor* 

Gartner, N., Messer, C. J., & Rathi, A. K. (2001). Traffic flow theory: A state-of-the-art report.

Gazis, D. C., Herman, R., & Potts, R. B. (1959). Car-following theory of steady-state traffic

Gazis, D. C., Herman, R., & Rothery, R. W. (1961). Nonlinear follow-the-leader models of

Gipps, P. G. (1981). A behavioural car-following model for computer simulation.

Gomes, G., May, A., & Horowitz, R. (2004). Congested freeway microsimulation model

Gonzalez-Rojo, S., Slama, J., Lopes, P. A., & Mora-Camino, F. (2002). A fuzzy logic approach for car-following modelling. *Systems Analysis Modelling Simulation, 42*(5), 735-755.

Greenshields, B. (1935b). A study in highway capacity. Highway Res. *Board Proc. v14*, 448-

Hall, F. L. (1996). Traffic stream characteristics. *Traffic Flow Theory. US Federal Highway* 

Herman, R., Montroll, E. W., Potts, R. B., & Rothery, R. W. (1959). Traffic dynamics: analysis

Herman, R., & Rothery, R. (1962). Microscopic and macroscopic aspects of single lane traffic

Jacob, C., & Abdulhai, B. (2006). Automated adaptive traffic corridor control using

reinforcement learning: approach and case studies. *Transportation Research Record:* 

using VISSIM. *Transportation Research Record: Journal of the Transportation Research* 

*Transportation Research Part B: Methodological, 15*(2), 105-111.

Greenberg, H. (1959). An analysis of traffic flow. *Operations Research*, 79-85. Greenshields, B. (1935a). *A study in highway capacity, highway research board*.

of stability in car following. *Operations Research*, 86-106.

*Journal of the Transportation Research Board, 1959*(-1), 1-8.

flow. *Operations Research, Japan*, 74.

Calculation of Traffic Analysis Tools Measures of Effectiveness: Federal Highway

*Manuals Never Told You*.

Highway Res: Record.

*different real-world situations*.

*Using Microscopic Simulation*.

*Board, 1876*(-1), 71-81.

477.

*Administration*.

flow. *Operations Research*, 499-505.

traffic flow. *Operations Research*, 545-567.

*Transportation Research Board, Washington DC*.

*Research*, 66-76.

of Engineering.

*Record*.

317-321.

Administration Report FHWA-HOP-08-054.


**5** 

Ardavan Rahimian *University of Birmingham* 

*United Kingdom* 

**Microwave Beamforming Networks for** 

An Intelligent Transportation System (ITS) is a system based on wireless communications which has been investigated for many years in order to provide new technologies able to improve safety and efficiency of road transportation with integrated vehicle and road systems. It combines all aspects of technology and systems design concepts in order to develop and improve transportation system of all kinds. ITS, which utilise information and communications technology in vehicle as well as within the roadside infrastructure, can also be used to improve mobility while increasing transport safety, reducing traffic congestion, maximising comfort, and reducing environmental impact (Andrisano et al., 2000). Intelligent transportation systems and applications can improve the quality of travel by selecting routes with up-to-the-minute information data, giving priority to response vehicle teams, notifying drivers about road incidents, and delivering ITS services to drivers. They can reduce fuel consumption by routing the vehicles to their destinations so that fuel waste is significantly reduced, and also fully utilise the capacity of the existing road vehicular networks by controlling the flow of vehicles based on traffic monitoring and detecting congestions.

Vehicles within the ITS framework have to work in an autonomous manner to sense the driving environment and in a cooperative manner to exchange information data such as braking and acceleration between vehicles and also traffic, road, and weather conditions between vehicles and roadside units (Han & Wu, 2011). Hence, radio communications links between vehicles on a motorway are envisaged, leading to the formation of ad-hoc networks between clusters of vehicles and roadside beacons. System performance and analysis can be improved in various ways by the use of smart antenna systems and techniques. These microwave systems fulfil the requirements of improving coverage range, capacity, data-rate, and quality of service (QoS). Smart antenna systems are generally classified as either switched-beam antenna systems or adaptive arrays. Switched-beam antenna systems use fixed multiple beamforming networks (BFNs) in order to create various beam patterns based on the different microwave beamforming techniques and technologies. These smart systems can be used to increase the wireless channel capacity limited by the presence of interference. By using narrow beams available from these systems, it is possible to increase the gain in the desired signal direction and to reduce it toward interference directions. They can also be used for mobile communication base stations in order to provide space-division multiple access (SDMA) capabilities. On the other hand, growing demand in intelligent transportation systems means there is a need for multiple antennas with multiple beams.

**1. Introduction** 

**Intelligent Transportation Systems** 


## **Microwave Beamforming Networks for Intelligent Transportation Systems**

Ardavan Rahimian *University of Birmingham United Kingdom* 

#### **1. Introduction**

122 Intelligent Transportation Systems

Payne, H. J. (1979). FREFLO: A macroscopic simulation model of freeway traffic.

Pipes, L. A. (1967). Car following models and the fundamental diagram of road traffic.

Prevedouros, P. D., & Wang, Y. (1999). Simulation of large freeway and arterial network

Rockwell, T. H., Ernest, R. L., & Hanken, A. (1968). A Sensitivity Analysis of Empirical Car-

Rouphail, N., Tarko, A., & Li, J. (1996). Traffic Flow at Signalized Intersections. in Traffic Flow Theory. *Washington, DC: US Federal Highway Administration*, 9-1. Troutbeck, R., & Brilon, W. (1997). Unsignalized Intersection Theory, Revised Traffic Flow

Underwood, R. (1961). Speed, volume, and density relationships: Quality and theory of

Van Winsum, W. (1999). The human element in car following models. *Transportation* 

Waldeer, K. T. (2004). Numerical investigation of a mesoscopic vehicular traffic flow model based on a stochastic acceleration process. *Arxiv preprint cond-mat/0412490*. Wicks, D., Lieberman, E. B., Associates, K., & Division, U. S. F. H. A. T. S. (1980).

*Final Report*: Federal Highway Administration, Traffic Systems Division. Wiedemann, R. (1974). Simulation of road traffic flow. *Reports of the Institute for Transport and* 

Williams, J. (1996). Macroscopic Flow Models in Traffic Flow Theory. *Washington, DC: US* 

Zhang, L., McHale, G., & Zhang, Y. (2003). Modeling and validating CORSIM freeway

origin-destination volumes. *Transportation Research Record: Journal of the* 

Williams, J. C. (1997). Macroscopic flow models. *Revised monograph on traffic flow theory*. Yikai, K., Satoh, J., Itakura, N., Honda, N., & Satoh, A. (1993). *A fuzzy model for behaviour of* 

*Development and Testing of INTRAS, a Microscopic Freeway Simulation Model: Vol. 1, Program Design, Parameter Calibration and Freeway Dynamics Component Development:* 

*Journal of the Transportation Research Board, 1678*(-1), 197-207.

Following Models. *Transportation Research, 2*, 363-373.

Rothery, R. W. (1992). Car following models. *Trac Flow Theory*.

traffic flow. *Yale bureau of highway traffic*, 141–188.

Waldeer, K. (2006). *Kinetic Theory in Vehicular Traffic Flow Modeling*.

*Communication, University of Karlsruhe*.

*Federal Highway Administration*, 6-1.

*vehicles to analyze traffic congestion*.

*Transportation Research Board, 1856*(-1), 135-142.

*Research Part F: Traffic Psychology and Behaviour, 2*(4), 207-211.

with CORSIM, INTEGRATION, and WATSim. *Transportation Research Record:* 

*Transportation Research Record*(722).

*Transportation Research, 1*(1), 21-29.

PTV (2005). VISSIM Version 4.10. *User Manual, March*.

Theory.

An Intelligent Transportation System (ITS) is a system based on wireless communications which has been investigated for many years in order to provide new technologies able to improve safety and efficiency of road transportation with integrated vehicle and road systems. It combines all aspects of technology and systems design concepts in order to develop and improve transportation system of all kinds. ITS, which utilise information and communications technology in vehicle as well as within the roadside infrastructure, can also be used to improve mobility while increasing transport safety, reducing traffic congestion, maximising comfort, and reducing environmental impact (Andrisano et al., 2000). Intelligent transportation systems and applications can improve the quality of travel by selecting routes with up-to-the-minute information data, giving priority to response vehicle teams, notifying drivers about road incidents, and delivering ITS services to drivers. They can reduce fuel consumption by routing the vehicles to their destinations so that fuel waste is significantly reduced, and also fully utilise the capacity of the existing road vehicular networks by controlling the flow of vehicles based on traffic monitoring and detecting congestions.

Vehicles within the ITS framework have to work in an autonomous manner to sense the driving environment and in a cooperative manner to exchange information data such as braking and acceleration between vehicles and also traffic, road, and weather conditions between vehicles and roadside units (Han & Wu, 2011). Hence, radio communications links between vehicles on a motorway are envisaged, leading to the formation of ad-hoc networks between clusters of vehicles and roadside beacons. System performance and analysis can be improved in various ways by the use of smart antenna systems and techniques. These microwave systems fulfil the requirements of improving coverage range, capacity, data-rate, and quality of service (QoS). Smart antenna systems are generally classified as either switched-beam antenna systems or adaptive arrays. Switched-beam antenna systems use fixed multiple beamforming networks (BFNs) in order to create various beam patterns based on the different microwave beamforming techniques and technologies. These smart systems can be used to increase the wireless channel capacity limited by the presence of interference. By using narrow beams available from these systems, it is possible to increase the gain in the desired signal direction and to reduce it toward interference directions. They can also be used for mobile communication base stations in order to provide space-division multiple access (SDMA) capabilities. On the other hand, growing demand in intelligent transportation systems means there is a need for multiple antennas with multiple beams.

Microwave Beamforming Networks for Intelligent Transportation Systems 125

equipment and processing power, however, the use of smart antennas is currently limited to

Smart antennas are also beneficial in multi-user vehicular scenarios, in order to suppress CCI. Again, both transmitter- and receiver-sided microwave beamforming can be employed for mitigating CCI. When transmitting, each user can adjust the beam pattern such that there are nulls in the directions of other co-channel users and a high directivity towards the desired direction of radiowave transmission. Hence, the SINR for the other co-channel users is improved as well as the SNR at the desired receiver. Similarly, when receiving each user can adjust the beam pattern such that directions of other co-channel interferers are nulled and desired directions of reception are enhanced and therefore each user can improve the received SINR. The use of smart antenna systems for CCI cancellation offers the opportunity to accommodate multiple co-channel users within the same frequency band. This concept is

Fig. 1. Rotman Lens Microwave BFN Configuration, taken from (Mietzner et al., 2009).

ports through a multiple-way RF switch giving a sequentially scanning antenna.

A multiple microwave beamforming network is one with a capability to form many beams in different directions from the same aperture. If a separate RF transmit or receive system is connected to each beam port, simultaneous independent operation in many directions can be obtained. Alternatively, a single transmit or receive system can be connected to the beam

Switched-beam smart antenna systems may be cheaper than an equivalent phased array, particularly when few beam signals are needed. The creation of a multiple beam antenna using a multiple microwave beamforming network has the advantage that no devices for frequency changing are necessary. The technique therefore has the potential to be simpler and lower in cost than IF, digital, and optical frequency methods. Indeed many antenna configurations, such as lenses, have inherent multiple beam capabilities. In these cases it is only necessary to replace the single feed by an array so that each array element forms one of the multiple beams. The field of microwave beamforming networks encompasses two main research areas namely lens-based quasi-optic types, involving a hybrid arrangement of either a lens objective with a feed array, and circuit-based types used to feed antenna arrays. Circuit-based microwave beamforming networks use transmission lines, connecting power splitters, and hybrid couplers in order to form multiple beam networks. The phase shifts required to produce multi-beam scanning are provided by lengths of transmission line. Lens-based microwave beamforming networks will produce high-gain beams over narrow scan ranges with lenses giving better beam control due to their increased design degrees of

stations that are fixed on vehicles (Mietzner et al., 2009).

referred to as SDMA (Mietzner et al., 2009).

Switched-beam antenna systems can greatly improve the performance of the intelligent transportation systems by providing better link quality and high immunity to interference.

Also, generating multiple beams using an array along with having wide bandwidth and beam steering capability are of crucial importance for modern wireless communication systems. For this purpose, various multiple beamforming networks are introduced to have control over the amplitude and phase at each element of the antenna array. Microwave passive networks form an important class of these networks and they have been widely used in switched-beam antenna systems. Two well-known examples of such networks are the Rotman Lens (lens-based beamforming network) and the Butler Matrix (circuit-based beamforming network). They increase the system capacity and provide higher signal-tointerference ratio, consequently enhancing the overall automotive telematics performance.

This chapter presents the novel designs of steerable microwave beamforming networks employing an 8×8 Rotman Lens for operation at 6.3 GHz (C-band), and cascaded 4×4 Butler Matrices for operation at 3.15 GHz (S-band). The microwave beamforming networks are intended for intelligent transportation systems and applications. Although the frequency range likely to be allocated to such systems is 63 GHz, where the short transmission range allows multiple frequency re-use, the microwave networks are frequency scaled models to verify the concept. The objective of this investigation is to develop microwave beamforming networks suitable for a use in vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. The microwave beamforming networks demonstrate appropriateness to develop well-established designs for systems that can be utilised in ITS applications and framework and vehicular ad-hoc network (VANET) telematics which is the convergence of telecommunications and information processing with application of vehicle tracking.

#### **2. Smart antennas and microwave beamforming techniques**

In addition to higher bit-rates and smaller error-rates, microwave beamforming techniques can also be utilised in order to improve the signal-to-noise ratio (SNR) at the receiver and to suppress co-channel interference (CCI) in a multi-user vehicular scenario, thus improving the SINR at the receiver. Using microwave beamforming techniques, the beam patterns of the antenna array can be steered in certain desired directions, whereas undesired directions can be suppressed. Consider an antenna array with *N* antenna elements, which receives an RF signal from a certain direction. Due to the geometry of the antenna array, the impinging RF signal reaches the individual antenna elements at different time instants, which causes phase shifts between the different received signals. If the direction of the impinging RF signal is known, the phase differences of the RF signals can be compensated by means of phase shifters or delay elements, before the received signals are added up. As a result, the overall antenna pattern of the phase array will exhibit a maximum in the direction of the impinging signal. This principle is called microwave beamforming and is shown in Fig. 1, which is equivalent to a mechanical rotation of the array. In a vehicular communication scenario, transmitted RF signals often propagate via a line-of-sight (LOS) path between transmitter and receiver and via paths that are associated with significant reflectors and diffractors in the environment (such as large trucks). If the directions of these dominant propagation paths are known at the receiver side, microwave beamforming can be applied in order to adjust the receiver beam pattern such that it has a high directivity towards the dominant angles of reception to accomplish significant antenna gains. Due to the required

Switched-beam antenna systems can greatly improve the performance of the intelligent transportation systems by providing better link quality and high immunity to interference. Also, generating multiple beams using an array along with having wide bandwidth and beam steering capability are of crucial importance for modern wireless communication systems. For this purpose, various multiple beamforming networks are introduced to have control over the amplitude and phase at each element of the antenna array. Microwave passive networks form an important class of these networks and they have been widely used in switched-beam antenna systems. Two well-known examples of such networks are the Rotman Lens (lens-based beamforming network) and the Butler Matrix (circuit-based beamforming network). They increase the system capacity and provide higher signal-tointerference ratio, consequently enhancing the overall automotive telematics performance. This chapter presents the novel designs of steerable microwave beamforming networks employing an 8×8 Rotman Lens for operation at 6.3 GHz (C-band), and cascaded 4×4 Butler Matrices for operation at 3.15 GHz (S-band). The microwave beamforming networks are intended for intelligent transportation systems and applications. Although the frequency range likely to be allocated to such systems is 63 GHz, where the short transmission range allows multiple frequency re-use, the microwave networks are frequency scaled models to verify the concept. The objective of this investigation is to develop microwave beamforming networks suitable for a use in vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. The microwave beamforming networks demonstrate appropriateness to develop well-established designs for systems that can be utilised in ITS applications and framework and vehicular ad-hoc network (VANET) telematics which is the convergence of

telecommunications and information processing with application of vehicle tracking.

In addition to higher bit-rates and smaller error-rates, microwave beamforming techniques can also be utilised in order to improve the signal-to-noise ratio (SNR) at the receiver and to suppress co-channel interference (CCI) in a multi-user vehicular scenario, thus improving the SINR at the receiver. Using microwave beamforming techniques, the beam patterns of the antenna array can be steered in certain desired directions, whereas undesired directions can be suppressed. Consider an antenna array with *N* antenna elements, which receives an RF signal from a certain direction. Due to the geometry of the antenna array, the impinging RF signal reaches the individual antenna elements at different time instants, which causes phase shifts between the different received signals. If the direction of the impinging RF signal is known, the phase differences of the RF signals can be compensated by means of phase shifters or delay elements, before the received signals are added up. As a result, the overall antenna pattern of the phase array will exhibit a maximum in the direction of the impinging signal. This principle is called microwave beamforming and is shown in Fig. 1, which is equivalent to a mechanical rotation of the array. In a vehicular communication scenario, transmitted RF signals often propagate via a line-of-sight (LOS) path between transmitter and receiver and via paths that are associated with significant reflectors and diffractors in the environment (such as large trucks). If the directions of these dominant propagation paths are known at the receiver side, microwave beamforming can be applied in order to adjust the receiver beam pattern such that it has a high directivity towards the dominant angles of reception to accomplish significant antenna gains. Due to the required

**2. Smart antennas and microwave beamforming techniques** 

equipment and processing power, however, the use of smart antennas is currently limited to stations that are fixed on vehicles (Mietzner et al., 2009).

Smart antennas are also beneficial in multi-user vehicular scenarios, in order to suppress CCI. Again, both transmitter- and receiver-sided microwave beamforming can be employed for mitigating CCI. When transmitting, each user can adjust the beam pattern such that there are nulls in the directions of other co-channel users and a high directivity towards the desired direction of radiowave transmission. Hence, the SINR for the other co-channel users is improved as well as the SNR at the desired receiver. Similarly, when receiving each user can adjust the beam pattern such that directions of other co-channel interferers are nulled and desired directions of reception are enhanced and therefore each user can improve the received SINR. The use of smart antenna systems for CCI cancellation offers the opportunity to accommodate multiple co-channel users within the same frequency band. This concept is referred to as SDMA (Mietzner et al., 2009).

Fig. 1. Rotman Lens Microwave BFN Configuration, taken from (Mietzner et al., 2009).

A multiple microwave beamforming network is one with a capability to form many beams in different directions from the same aperture. If a separate RF transmit or receive system is connected to each beam port, simultaneous independent operation in many directions can be obtained. Alternatively, a single transmit or receive system can be connected to the beam ports through a multiple-way RF switch giving a sequentially scanning antenna.

Switched-beam smart antenna systems may be cheaper than an equivalent phased array, particularly when few beam signals are needed. The creation of a multiple beam antenna using a multiple microwave beamforming network has the advantage that no devices for frequency changing are necessary. The technique therefore has the potential to be simpler and lower in cost than IF, digital, and optical frequency methods. Indeed many antenna configurations, such as lenses, have inherent multiple beam capabilities. In these cases it is only necessary to replace the single feed by an array so that each array element forms one of the multiple beams. The field of microwave beamforming networks encompasses two main research areas namely lens-based quasi-optic types, involving a hybrid arrangement of either a lens objective with a feed array, and circuit-based types used to feed antenna arrays. Circuit-based microwave beamforming networks use transmission lines, connecting power splitters, and hybrid couplers in order to form multiple beam networks. The phase shifts required to produce multi-beam scanning are provided by lengths of transmission line. Lens-based microwave beamforming networks will produce high-gain beams over narrow scan ranges with lenses giving better beam control due to their increased design degrees of

Microwave Beamforming Networks for Intelligent Transportation Systems 127

 *F1P* + *W*(*N*) + *N* sin*α* = *F* + *W*(0) (1)

 *F2P* + *W*(*N*) – *N* sin*α* = *F* + *W*(0) (2)

(*F1P*)2 = (*F* cos*α* + *X*)2 + (*F* sin*α* – *Y*)2

(*F2P*)2 = (*F* cos*α* + *X*)2 + (*F* sin*α* + *Y*)2

(*F0P*)2 = (*G* + *X*)2 + (*Y*)2

 *G* = *G* / *F* (5)

 *x*,*y* = *X*,*Y* / *F* (6)

 *y* = *η* (1 – *w*) (8)

 *aw*2 + *bw* + *c* = 0 (10)

In order to calculate performance of the microwave lens, the coupling between ports is approximated using aperture theory and a uniform distribution to the port aperture is implied. These port radiation patterns are used to compute the direct path and reflected path propagation from port to port. Also, to improve the response of the outer beams, the beam and array ports are adjusted so that each line is pointing toward the centre of the lens on the opposite side rather than being normal to the microwave lens surface (Maybell, 1981). Phase error is calculated by comparing electrical lengths along two distinct paths from a given beam port through the microwave lens (Fig. 3). The first path travels through any one of the off-axis array ports, through its taper and transmission line, and finally along the path from the array element phase centre to the beam phase front. The second path begins at the same beam port but travels through the centre of the array curve and through a length of line common to all array ports. The comparison of these electrical lengths obtains the phase error for this beam port. This is done over the list of beam ports to produce a phase error.

*WN W* ( )- (0) *<sup>w</sup>*

Manipulation of these equations leads to the following relations for *x*, *y*, and *w*:

The lens design program solves for these points each time *F*, *η*, *α*, *g* are modified.

**3.2 Rotman lens microwave BFN performance and phase error analysis** 

 *x*2 + *y*2 + 2*a0x* = *w*2 + *b0*

Lens dimensions are then normalised by the off-axis focal length (*F*).

where

*F0P* + *W*(*N*) = *G* + *W*(0) (3)

*η* = *N* / *F* (4)

*<sup>F</sup>* (7)

<sup>2</sup>*η*2 – 2*w* (9)

freedom. Circuit-based networks (Butler Matrix) have the travelling wave or corporate feed characteristics and can be used in limited size arrays as can the Rotman Lens, which in addition will give wide bandwidth (Hall & Vetterlein, 1990).

#### **3. Rotman lens microwave BFN analysis and design**

The Rotman Lens is an attractive passive microwave lens-based beamforming network due to its low cost, reliability, design simplicity, and wide-angle scanning capabilities. It is a device that uses the free-space wavelength of a signal injected into a geometrically configured waveguide to passively shift the phase of inputs into a linear antenna array in order to scan a beam in any desired signal pattern. It has a carefully chosen shape and appropriate length transmission lines in order to produce a wave-front across the output that is phased by the time-delay in the signal transmission.

A Rotman Lens achieves beam scanning using equivalent time-delays that are created by the different path lengths to the radiating elements. These lengths depend on the relative position between the beam port and the array ports on the structure. As long as the path lengths exhibit constant time-delay behaviour over the bandwidth, the lens is insensitive to the beam squint problems exhibited by constant phase beamformers (Weiss, 2010). Each input port will produce a distinct beam that is shifted in angle at the system output. The design of the lens is controlled by a series of equations that set the focal points and array positions. The inputs, during the design of the system, include the desired number of beams and array elements, and the spacing of the elements (Penney, 2008). Since Rotman Lens is a true-time-delay (TTD) device, it produces beam steering independent of frequency and is therefore capable of wide-band operation. Also, the cost of a Rotman Lens implemented on microwave material primarily driven by the cost of the material itself and the price of photo etching (Weiss et al., 2007).

#### **3.1 Rotman lens microwave BFN contour synthesis**

The synthesis of the microwave lens assumes several input parameters which are used to compute the inner contour (array contour) point as well as the line lengths. These parameters are element spacing (*η*), focal ratio (*g*), lens width (*G* or *F0*), and scan angle (*α*) (Fig. 2). The lens inner contour points and transmission line lengths are solved for using the technique of path length comparison (Rotman & Turner, 1963).

Fig. 2. Rotman Lens Microwave BFN Configuration.

$$F\_1 P + \mathcal{W}(\mathcal{N}) + \mathcal{N} \sin a = F + \mathcal{W}(\mathcal{O}) \tag{1}$$

$$F\_2P + \mathcal{W}(\mathcal{N}) - \mathcal{N}\sin a = F + \mathcal{W}(0) \tag{2}$$

$$F\_0 P + \mathcal{V}\mathcal{V}(\mathcal{N}) = G + \mathcal{V}\mathcal{V}(\mathcal{O}) \tag{3}$$

where

126 Intelligent Transportation Systems

freedom. Circuit-based networks (Butler Matrix) have the travelling wave or corporate feed characteristics and can be used in limited size arrays as can the Rotman Lens, which in

The Rotman Lens is an attractive passive microwave lens-based beamforming network due to its low cost, reliability, design simplicity, and wide-angle scanning capabilities. It is a device that uses the free-space wavelength of a signal injected into a geometrically configured waveguide to passively shift the phase of inputs into a linear antenna array in order to scan a beam in any desired signal pattern. It has a carefully chosen shape and appropriate length transmission lines in order to produce a wave-front across the output

A Rotman Lens achieves beam scanning using equivalent time-delays that are created by the different path lengths to the radiating elements. These lengths depend on the relative position between the beam port and the array ports on the structure. As long as the path lengths exhibit constant time-delay behaviour over the bandwidth, the lens is insensitive to the beam squint problems exhibited by constant phase beamformers (Weiss, 2010). Each input port will produce a distinct beam that is shifted in angle at the system output. The design of the lens is controlled by a series of equations that set the focal points and array positions. The inputs, during the design of the system, include the desired number of beams and array elements, and the spacing of the elements (Penney, 2008). Since Rotman Lens is a true-time-delay (TTD) device, it produces beam steering independent of frequency and is therefore capable of wide-band operation. Also, the cost of a Rotman Lens implemented on microwave material primarily driven by the cost of the material itself and the price of photo

The synthesis of the microwave lens assumes several input parameters which are used to compute the inner contour (array contour) point as well as the line lengths. These parameters are element spacing (*η*), focal ratio (*g*), lens width (*G* or *F0*), and scan angle (*α*) (Fig. 2). The lens inner contour points and transmission line lengths are solved for using the

addition will give wide bandwidth (Hall & Vetterlein, 1990).

**3. Rotman lens microwave BFN analysis and design** 

that is phased by the time-delay in the signal transmission.

**3.1 Rotman lens microwave BFN contour synthesis** 

Fig. 2. Rotman Lens Microwave BFN Configuration.

technique of path length comparison (Rotman & Turner, 1963).

etching (Weiss et al., 2007).

$$(F\_1P)^2 = (F\cos a + X)^2 + (F\sin a - Y)^2$$

$$(F\_2P)^2 = (F\cos a + X)^2 + (F\sin a + Y)^2$$

$$(F\_0P)^2 = (G+X)^2 + (Y)^2$$

Lens dimensions are then normalised by the off-axis focal length (*F*).

$$
\eta = \mathcal{N} \;/\; F \tag{4}
$$

$$\mathbf{G} = \mathbf{G} \nmid F \tag{5}$$

$$\mathbf{x}, \mathbf{y} = \mathbf{X}, \mathbf{Y} \;/\; F \tag{6}$$

$$w = \frac{\mathcal{W}(\mathcal{N}) \cdot \mathcal{W}(0)}{F} \tag{7}$$

Manipulation of these equations leads to the following relations for *x*, *y*, and *w*:

$$y = \eta \begin{pmatrix} 1 \ -w \end{pmatrix} \tag{8}$$

$$x^2 + y^2 + 2a\chi = w^2 + bv^2\eta^2 - 2w$$

$$aw^2 + bw + c = 0\tag{10}$$

The lens design program solves for these points each time *F*, *η*, *α*, *g* are modified.

#### **3.2 Rotman lens microwave BFN performance and phase error analysis**

In order to calculate performance of the microwave lens, the coupling between ports is approximated using aperture theory and a uniform distribution to the port aperture is implied. These port radiation patterns are used to compute the direct path and reflected path propagation from port to port. Also, to improve the response of the outer beams, the beam and array ports are adjusted so that each line is pointing toward the centre of the lens on the opposite side rather than being normal to the microwave lens surface (Maybell, 1981).

Phase error is calculated by comparing electrical lengths along two distinct paths from a given beam port through the microwave lens (Fig. 3). The first path travels through any one of the off-axis array ports, through its taper and transmission line, and finally along the path from the array element phase centre to the beam phase front. The second path begins at the same beam port but travels through the centre of the array curve and through a length of line common to all array ports. The comparison of these electrical lengths obtains the phase error for this beam port. This is done over the list of beam ports to produce a phase error.

Microwave Beamforming Networks for Intelligent Transportation Systems 129

The above microwave lens has an elliptical curvature on the beam port side. In this design, dummy ports are replaced with the terminated absorber sidewalls in order to introduce a novelty in the microwave lens structure, reduce the network size and unwanted reflections (and therefore reduce phase errors at the array ports), and increase the performance of the lens. The geometry of the transmission line routing is adjusted in a way to ensure no overlapping, proper spacing between lines, proper curvature, and maintaining overall lens physical length requirement. To obtain the desired performance, the lens requires to be tuned in terms of phase error or the array factor. The tuning involves adjusting the focal ratio (*g*) of the lens that will minimise the error reported by the phase error. This factor determines the curvature and focus of the lens, and if not adjusted accurately, will produce a messy beam. Hence, the focal ratio (*g*) is adjusted to 1.2670 in order to minimise the beam

Fig. 4. 8×8 ITS Rotman Lens Microwave BFN Configuration.

to array phase error and to produce well-focused beams (Fig. 5).

Fig. 5. 8×8 ITS Rotman Lens Microwave BFN Beam to Array Phase Error.

The simulated results for the 8×8 ITS Rotman Lens steerable beamforming network indicate the expected outcomes by having the main lobe of the array factor radiation pattern more than 10 dB greater than the side lobes and having a linear phase shift at each output port as

Phase Error Δφ = Rb – Ra (12)

where

$$R\_b = \begin{array}{c} \lfloor R\_b \rfloor + qr\_{TL} + y \text{ ( $n$ )} \sin \text{ ( $a$ )}\\\\ R\_a = \begin{array}{c} \lfloor R\_a \rfloor + qr\_{TLO} \end{array} \end{array}$$

Fig. 3. Rotman Lens Microwave BFN Phase Error Calculation.

When a feed point is placed at any one of the focal points, the corresponding wave-front has no phase error. When the feed is displaced from these lens focal points, the corresponding wave-front will have a phase error. However, for wide-angle microwave beam scanning, the lens must be focused at all intermediate points along the focal arc.

#### **3.3 8×8 ITS Rotman lens microwave BFN design and performance**

As Fig. 4 indicates, a realistic 8×8 ITS Rotman Lens microwave beamforming network is designed and simulated for a use in intelligent transportation systems and applications. The design parameters are based on those used in previous section. In this case, the microwave lens is designed to have 8 beam ports, 8 array ports suitable for an 8-element antenna array, a beam scan angle of ±50° at a centre frequency of 6.3 GHz, and an element spacing of 28 mm. The prototype for the lens is fabricated as a microstrip with a 50 Ω impedance transmission lines on Taconic TLC-30 substrate with the dielectric constant (*εr*) of 3.0, substrate thickness (H) of 1.3 mm, and loss tangent of 0.003. The design gives the microwave lens a compact size of 35.91 cm × 25.80 cm. The array ports have also been spaced in such a way that elements of the antenna array can be directly attached to the microwave network. Design, synthesis, and analysis of the 8×8 microwave Rotman Lens and its variants are based on real-time analysis of geometrical optics (GO).

*Rb* = |*Rb*| + *φTLi* + *y* (*n*) sin (*αi*)

*Ra* = |*Ra*| + *φTL0* 

When a feed point is placed at any one of the focal points, the corresponding wave-front has no phase error. When the feed is displaced from these lens focal points, the corresponding wave-front will have a phase error. However, for wide-angle microwave beam scanning, the

As Fig. 4 indicates, a realistic 8×8 ITS Rotman Lens microwave beamforming network is designed and simulated for a use in intelligent transportation systems and applications. The design parameters are based on those used in previous section. In this case, the microwave lens is designed to have 8 beam ports, 8 array ports suitable for an 8-element antenna array, a beam scan angle of ±50° at a centre frequency of 6.3 GHz, and an element spacing of 28 mm. The prototype for the lens is fabricated as a microstrip with a 50 Ω impedance transmission lines on Taconic TLC-30 substrate with the dielectric constant (*εr*) of 3.0, substrate thickness (H) of 1.3 mm, and loss tangent of 0.003. The design gives the microwave lens a compact size of 35.91 cm × 25.80 cm. The array ports have also been spaced in such a way that elements of the antenna array can be directly attached to the microwave network. Design, synthesis, and analysis of the 8×8 microwave Rotman Lens and its variants are

Fig. 3. Rotman Lens Microwave BFN Phase Error Calculation.

lens must be focused at all intermediate points along the focal arc.

based on real-time analysis of geometrical optics (GO).

**3.3 8×8 ITS Rotman lens microwave BFN design and performance** 

where

Phase Error Δφ = Rb – Ra (12)

Fig. 4. 8×8 ITS Rotman Lens Microwave BFN Configuration.

The above microwave lens has an elliptical curvature on the beam port side. In this design, dummy ports are replaced with the terminated absorber sidewalls in order to introduce a novelty in the microwave lens structure, reduce the network size and unwanted reflections (and therefore reduce phase errors at the array ports), and increase the performance of the lens. The geometry of the transmission line routing is adjusted in a way to ensure no overlapping, proper spacing between lines, proper curvature, and maintaining overall lens physical length requirement. To obtain the desired performance, the lens requires to be tuned in terms of phase error or the array factor. The tuning involves adjusting the focal ratio (*g*) of the lens that will minimise the error reported by the phase error. This factor determines the curvature and focus of the lens, and if not adjusted accurately, will produce a messy beam. Hence, the focal ratio (*g*) is adjusted to 1.2670 in order to minimise the beam to array phase error and to produce well-focused beams (Fig. 5).

Fig. 5. 8×8 ITS Rotman Lens Microwave BFN Beam to Array Phase Error.

The simulated results for the 8×8 ITS Rotman Lens steerable beamforming network indicate the expected outcomes by having the main lobe of the array factor radiation pattern more than 10 dB greater than the side lobes and having a linear phase shift at each output port as

Microwave Beamforming Networks for Intelligent Transportation Systems 131

As Fig. 7a and Fig. 7b indicate, the array factor radiation patterns for ports 1 to 8 have their beams formed in the expected signal directions having their main lobes showed strong identity with at least 10 dB of isolation from the side lobes. It is shown that despite the nonideal performance of microwave network, in terms of phase and amplitude distributions, it

Fig. 7. a. 8×8 ITS Rotman Lens Microwave BFN Measured Array Factor Radiation Patterns for Port 1 to Port 4. b. 8×8 ITS Rotman Lens Microwave BFN Measured Array Factor

Radiation Patterns for Port 5 to Port 8.

b

a

a function of frequency. The beam to array coupling amplitude (the array ports distribution from a given beam or set of beams) for the array ports 9 to 16 has expected outcome of –9 dB to –13 dB has been obtained as a result of the accurate lens design and it confirms how the amplitude distribution along the array contour is much more uniform with beam port pointing enabled. Also, the progressive phase shift for the lens array ports exciting beam port 5 ensures the generation of eight distinct beams and beam scanning capabilities. It is computed using the linear distance between the ports in the dielectric medium chosen.

The fabrication of microwave network was carried out and the lens was then extensively measured on a network analyser over a frequency range of 5.5 GHz to 7.5 GHz with the frequency of operation as 6.3 GHz. Fig. 6 indicates the fabricated microwave network being tested. Only one beam port and one array port (*S21*) are measured at a time and all other ports are perfectly terminated using 50 Ω termination loads.

Fig. 6. Fabricated 8×8 ITS Rotman Lens Microwave BFN.

The array factor (*AF*) is a function of the geometry of the antenna array element and the excitation phase and it quantifies the effect of combining radiating elements in an array without the element specific radiation pattern taken into account. The array factor of an *N*element antenna array is given by (Raisanen & Lehto, 2003):

$$\begin{aligned} \{AF\} &= \sum\_{n=1}^{M} a\_n \cos[(2n-1)\mu] \\ \text{where,} &\mu = \frac{\pi d}{\lambda} \cos \theta \end{aligned} \tag{13}$$

where *d* is the separation between the elements, *M* is the number of isotropic elements, and *an*'s are the excitation coefficients of the array elements. By substituting the values to (13), the array factor radiation patterns for the proposed microwave beamforming network can be computed to verify the ITS beam beam scanning electronically steered arrays concept.

a function of frequency. The beam to array coupling amplitude (the array ports distribution from a given beam or set of beams) for the array ports 9 to 16 has expected outcome of –9 dB to –13 dB has been obtained as a result of the accurate lens design and it confirms how the amplitude distribution along the array contour is much more uniform with beam port pointing enabled. Also, the progressive phase shift for the lens array ports exciting beam port 5 ensures the generation of eight distinct beams and beam scanning capabilities. It is computed using the linear distance between the ports in the dielectric medium chosen.

The fabrication of microwave network was carried out and the lens was then extensively measured on a network analyser over a frequency range of 5.5 GHz to 7.5 GHz with the frequency of operation as 6.3 GHz. Fig. 6 indicates the fabricated microwave network being tested. Only one beam port and one array port (*S21*) are measured at a time and all other

The array factor (*AF*) is a function of the geometry of the antenna array element and the excitation phase and it quantifies the effect of combining radiating elements in an array without the element specific radiation pattern taken into account. The array factor of an *N*-

> 1 ( ) cos[(2 -1) ]

 

*M n n*

*<sup>d</sup> where u* 

, cos

where *d* is the separation between the elements, *M* is the number of isotropic elements, and *an*'s are the excitation coefficients of the array elements. By substituting the values to (13), the array factor radiation patterns for the proposed microwave beamforming network can be computed to verify the ITS beam beam scanning electronically steered arrays concept.

(13)

*AF a n u*

ports are perfectly terminated using 50 Ω termination loads.

Fig. 6. Fabricated 8×8 ITS Rotman Lens Microwave BFN.

element antenna array is given by (Raisanen & Lehto, 2003):

As Fig. 7a and Fig. 7b indicate, the array factor radiation patterns for ports 1 to 8 have their beams formed in the expected signal directions having their main lobes showed strong identity with at least 10 dB of isolation from the side lobes. It is shown that despite the nonideal performance of microwave network, in terms of phase and amplitude distributions, it

Fig. 7. a. 8×8 ITS Rotman Lens Microwave BFN Measured Array Factor Radiation Patterns for Port 1 to Port 4. b. 8×8 ITS Rotman Lens Microwave BFN Measured Array Factor Radiation Patterns for Port 5 to Port 8.

Microwave Beamforming Networks for Intelligent Transportation Systems 133

In this ITS microwave beamforming network, four hybrid couplers, two crossovers, and two fixed phase shifters are combined to obtain the 4×4 Butler Matrix. The phase differences are ±45° and ±135° from port 1 and port 4, and port 2 and port 3, respectively. The output ports have been spaced in such a way that elements of the antenna array can be directly attached to the microwave network. If the matrix is connected to an antenna array, then the network will act so that the array will have a uniform amplitude distribution and constant phase difference between adjacent elements to generate orthogonal beams. The Butler Matrices are then cascaded in order to produce narrow-beam and broad-beam output that could provide multi-channel operation for automotive telematics applications, particularly for vehicle-to-

The Butler Matrix microwave beamforming network is theoretically lossless in that no power is intentionally dissipated in terminations. There will always be, however, a finite insertion loss due to the inherent losses in the hybrid couplers, fixed phase shifters, and transmission lines that make up the matrix. The Butler Matrix passive beamforming antenna also requires that the individual beam patterns be orthogonal in space (Skolnik, 2000).

Independent orthogonal beams mean that when two or more beam input ports are simultaneously excited, the resulting radiation is a linear superposition of the radiations that would be obtained when the ports are excited separately. In addition, when a signal is applied to one port it should have no output at the other ports. An antenna which is lossless and passive means that the radiated power is the same as the input power. Fig. 8 shows the topology of the Butler Matrix. The phase shift at the matrix output ports can be determined by summing up all the phase shifts of signal paths. Table 1 also indicates the resulting phase shift's characteristics at the matrix output elements. It was designed in such a way that when current excited to any input ports will only has one constant as shown in Table 1.

Fig. 8. Topology and Routing of Signal Paths of the 4×4 Butler Matrix Microwave BFN.

*Input Port 2*

*Port 5* –45° –135° –90° –180°

*Port 6* –90° 0° –225° –135°

*Port 7* –135° –225° 0° –90°

*Port 8* –180° –90° –135° –45° *Δ Phase* –45° +135° –135° +45° Table 1. Phase Shift's Characteristics at the Output Ports of the 4×4 Butler Matrix BFN.

*Input Port 3*

*Input Port 4*

*Port Input*

*Output*

*Output*

*Output*

*Output*

*Port 1*

vehicle (V2V) and vehicle-to-infrastructure (V2I) automotive communications.

is still capable of forming well-defined beams suitable for the beam steering experiments by causing the main lobe to be directed in certain directions for ports 1 to 8.

The difference in signal beam shape between the measured radiation patterns and simulated array factors is mainly in the nulls between the beams, which are not deep enough as the measured results because of small phase and amplitude deviations, the cross-coupling effects that are not taken into account in simulations, non-uniformity of transmission line width, and errors occurred during fabrication, measurement, and soldering. By eliminating the mentioned errors and using a shielded metal box with absorbing foams attached to the inner lid to reduce the interferences, the overall system performance will be improved and enhanced in terms of achieving high-gain narrow-beams with desired directions, and the relative phase shift will have a uniform distribution.

The proposed system can be used as the radio zone control technology units which scans the radio zone (antenna beam) in accordance with the average speed of a vehicle group in order to decrease the number of handovers within the specified continuous area. The microwave lens can also be integrated with amplifiers between the lens and the radiating elements as well as an RF switch array for selection of the signal beam ports and an A/D converter which samples the received signal and converts it to a digital signal and a DSP processor unit which then performs a Fast Fourier Transform (FFT) of the digital signal, and the amplitude and the phase parts are separated out.

ITS applications have generally been classified into three main categories with respect to their functionalities as safety, efficiency, and comfort applications. Safety applications minimise the risk of accidents and reduce the severity of the accident if it still occurs (collision avoidance, road sign notifications, incident management). Efficiency applications increase traffic efficiency by managing the traffic flow, and monitoring vehicles and road conditions. The purpose of comfort applications is to provide entertainment facilities and information to passengers by means of Internet access technologies (Dar et al., 2010).

Integration of digital signal processing unit with microwave beamforming network based on the 8×8 Rotman Lens will form a hybrid microwave-digital distributed beamforming network that can be employed in vehicular phased arrays and collision avoidance radar systems in order to support the ITS safety applications. Also, the proposed microwave network system can further be extended to wide-band structures to support the frequency of operation of 63 GHz for vehicle communication systems.

#### **4. Butler matrix microwave BFN analysis and design**

The Butler Matrix which is recently used due to its easy fabrication process and low cost, is a method of feeding an antenna array. It requires *N* beam (input) ports, *N* output ports, (*N*/2) log2(*N*) hybrid couplers, and (*N*/2) log2(*N*–1) fixed phase shifters to form the *N*×*N* network for an *N*-element array (Ahmad & Seman, 2005). When a signal excites an input beam port of the matrix, it produces different inter-element phase shifts between the output ports. To calculate the number of crossovers needed (14) may be used (Corona & Lancaster, 2003):

$$\mathbf{C}\_{p} = \mathbf{2}\mathbf{C}\_{p-1} + \mathbf{2}^{p-2}(\mathbf{2}^{p-2} - 1)\tag{14}$$

where *p* is the matrix order, which is related to the number of ports by *N*=2*p*. In equation (14), *p* should be equal or greater than 2 and *C1* = 1.

is still capable of forming well-defined beams suitable for the beam steering experiments by

The difference in signal beam shape between the measured radiation patterns and simulated array factors is mainly in the nulls between the beams, which are not deep enough as the measured results because of small phase and amplitude deviations, the cross-coupling effects that are not taken into account in simulations, non-uniformity of transmission line width, and errors occurred during fabrication, measurement, and soldering. By eliminating the mentioned errors and using a shielded metal box with absorbing foams attached to the inner lid to reduce the interferences, the overall system performance will be improved and enhanced in terms of achieving high-gain narrow-beams with desired directions, and the

The proposed system can be used as the radio zone control technology units which scans the radio zone (antenna beam) in accordance with the average speed of a vehicle group in order to decrease the number of handovers within the specified continuous area. The microwave lens can also be integrated with amplifiers between the lens and the radiating elements as well as an RF switch array for selection of the signal beam ports and an A/D converter which samples the received signal and converts it to a digital signal and a DSP processor unit which then performs a Fast Fourier Transform (FFT) of the digital signal, and the

ITS applications have generally been classified into three main categories with respect to their functionalities as safety, efficiency, and comfort applications. Safety applications minimise the risk of accidents and reduce the severity of the accident if it still occurs (collision avoidance, road sign notifications, incident management). Efficiency applications increase traffic efficiency by managing the traffic flow, and monitoring vehicles and road conditions. The purpose of comfort applications is to provide entertainment facilities and

Integration of digital signal processing unit with microwave beamforming network based on the 8×8 Rotman Lens will form a hybrid microwave-digital distributed beamforming network that can be employed in vehicular phased arrays and collision avoidance radar systems in order to support the ITS safety applications. Also, the proposed microwave network system can further be extended to wide-band structures to support the frequency

The Butler Matrix which is recently used due to its easy fabrication process and low cost, is a method of feeding an antenna array. It requires *N* beam (input) ports, *N* output ports, (*N*/2) log2(*N*) hybrid couplers, and (*N*/2) log2(*N*–1) fixed phase shifters to form the *N*×*N* network for an *N*-element array (Ahmad & Seman, 2005). When a signal excites an input beam port of the matrix, it produces different inter-element phase shifts between the output ports. To calculate the number of crossovers needed (14) may be used (Corona & Lancaster, 2003):

 *Cp* = 2*Cp–1* + 2*p–2*(2*p–2* – 1) (14) where *p* is the matrix order, which is related to the number of ports by *N*=2*p*. In equation

information to passengers by means of Internet access technologies (Dar et al., 2010).

causing the main lobe to be directed in certain directions for ports 1 to 8.

relative phase shift will have a uniform distribution.

amplitude and the phase parts are separated out.

of operation of 63 GHz for vehicle communication systems.

(14), *p* should be equal or greater than 2 and *C1* = 1.

**4. Butler matrix microwave BFN analysis and design** 

In this ITS microwave beamforming network, four hybrid couplers, two crossovers, and two fixed phase shifters are combined to obtain the 4×4 Butler Matrix. The phase differences are ±45° and ±135° from port 1 and port 4, and port 2 and port 3, respectively. The output ports have been spaced in such a way that elements of the antenna array can be directly attached to the microwave network. If the matrix is connected to an antenna array, then the network will act so that the array will have a uniform amplitude distribution and constant phase difference between adjacent elements to generate orthogonal beams. The Butler Matrices are then cascaded in order to produce narrow-beam and broad-beam output that could provide multi-channel operation for automotive telematics applications, particularly for vehicle-tovehicle (V2V) and vehicle-to-infrastructure (V2I) automotive communications.

The Butler Matrix microwave beamforming network is theoretically lossless in that no power is intentionally dissipated in terminations. There will always be, however, a finite insertion loss due to the inherent losses in the hybrid couplers, fixed phase shifters, and transmission lines that make up the matrix. The Butler Matrix passive beamforming antenna also requires that the individual beam patterns be orthogonal in space (Skolnik, 2000).

Independent orthogonal beams mean that when two or more beam input ports are simultaneously excited, the resulting radiation is a linear superposition of the radiations that would be obtained when the ports are excited separately. In addition, when a signal is applied to one port it should have no output at the other ports. An antenna which is lossless and passive means that the radiated power is the same as the input power. Fig. 8 shows the topology of the Butler Matrix. The phase shift at the matrix output ports can be determined by summing up all the phase shifts of signal paths. Table 1 also indicates the resulting phase shift's characteristics at the matrix output elements. It was designed in such a way that when current excited to any input ports will only has one constant as shown in Table 1.

Fig. 8. Topology and Routing of Signal Paths of the 4×4 Butler Matrix Microwave BFN.


Table 1. Phase Shift's Characteristics at the Output Ports of the 4×4 Butler Matrix BFN.

Microwave Beamforming Networks for Intelligent Transportation Systems 135

Fig. 11 and Fig. 13 indicate the simulated and measured S-parameters exciting beam port 1 and beam port 2 respectively. At an operating frequency of 3.15 GHz, the simulated results agree with the measured results. As Fig. 12 and Fig. 14 indicate, the measured progressive phase shift for the Butler Matrix microwave beamforming network output ports exciting port 1 and port 2 respectively ensure the generation of four different beams at 3.15 GHz.

By using the narrow-beam signals available from the ITS Butler Matrix, it is possible for a vehicle to increase gain in the desired signal directions and reduce the gain in interference signal directions. The differences between the simulations and measurements and slight distortion of the beam shape might be due to the employed FR4 board and fabrication process, non-uniformity of matrix transmission line width, cross-coupling effects, and measurement errors. This Butler Matrix microwave network can be used as a planar passive

The microwave beamforming network was designed to be placed anywhere on the envelope of the symmetrical cut-plane running through the centre of the vehicle. Possible antenna placement positions are therefore on the roof of the vehicle or inside the front and rear bumpers with a plastic radome. Antenna location is important to permit a mounting, which has little impact on vehicle styling, be of low cost, and be capable of addition to the vehicle

<sup>2500</sup> <sup>2600</sup> <sup>2700</sup> <sup>2800</sup> <sup>2900</sup> <sup>3000</sup> <sup>3100</sup> <sup>3200</sup> <sup>3300</sup> <sup>3400</sup> <sup>3500</sup> �80

Fig. 11. 4×4 ITS Butler Matrix Microwave BFN Simulated and Measured S-Parameters

Frequency (MHz)

 Measured S51 Measured S61 Measured S71 Measured S81 Simulated S51 Simulated S61 Simulated S71 Simulated S81

BM Port 1 Insertion Loss

BFN for multi-beam antennas used in automotive telematics and ITS applications.

with minimum re-design of surrounding components.

Fig. 10. Fabricated 4×4 ITS Butler Matrix Microwave BFN.

�70 �60 �50 �40 �30 �20 �10 0

Exciting Beam Port 1.

dB Insetion Loss

Fig. 9 shows the proposed planar configuration of the 4×4 ITS Butler Matrix microwave beamforming network obtained as a result of the accurate branch line coupler, crossover, and phase shifter components design. The input and output ports are connected through the phase shifters and branch line couplers such that when a signal is applied to any input port, the matrix produces equal amplitude signals at all the output ports. The 45 degree and 0 degree phase shifters, together with phase adjustment, are obtained by connecting a transmission line at the output port of the hybrid coupler to the input port of the other hybrid coupler. At the Butler Matrix output ports, additional transmission lines are placed in such a way that antenna array elements can be directly connected to the network. The design gives the 4×4 ITS Butler Matrix network a compact size of 11.6 cm × 9.1 cm for enhanced operation and better performance (Rahimian & Rahimian, 2010).

Fig. 9. Layout of the 4×4 ITS Butler Matrix Microwave BFN with Components.

#### **4.1 4×4 ITS Butler matrix microwave BFN realisation**

The fabrication of the 4×4 ITS Butler Matrix microwave beamforming network has been carried out and the measured results have slight error compared to simulated results. The prototype for the matrix is fabricated as a microstrip with a 50 Ω impedance transmission lines on FR4 substrate with the dielectric constant (*εr*) of 4.7 and thickness (H) of 0.8 mm, and loss tangent of 0.01. The ITS Butler Matrix microwave beamforming network was then extensively measured on a network analyser over a frequency range of 2.5 GHz to 3.5 GHz with the frequency of operation as 3.15 GHz. The Butler Matrix has been shielded with a metal box along with absorbing foam attached to the top lid of the inner box in order to reduce internal coupling and external interference effects. Fig. 10 indicates the fabricated ITS beamforming network being tested. Only one beam port and one output port are measured at a time and all other ports are perfectly terminated using 50 Ω termination loads.

Fig. 9 shows the proposed planar configuration of the 4×4 ITS Butler Matrix microwave beamforming network obtained as a result of the accurate branch line coupler, crossover, and phase shifter components design. The input and output ports are connected through the phase shifters and branch line couplers such that when a signal is applied to any input port, the matrix produces equal amplitude signals at all the output ports. The 45 degree and 0 degree phase shifters, together with phase adjustment, are obtained by connecting a transmission line at the output port of the hybrid coupler to the input port of the other hybrid coupler. At the Butler Matrix output ports, additional transmission lines are placed in such a way that antenna array elements can be directly connected to the network. The design gives the 4×4 ITS Butler Matrix network a compact size of 11.6 cm × 9.1 cm for

enhanced operation and better performance (Rahimian & Rahimian, 2010).

Fig. 9. Layout of the 4×4 ITS Butler Matrix Microwave BFN with Components.

at a time and all other ports are perfectly terminated using 50 Ω termination loads.

The fabrication of the 4×4 ITS Butler Matrix microwave beamforming network has been carried out and the measured results have slight error compared to simulated results. The prototype for the matrix is fabricated as a microstrip with a 50 Ω impedance transmission lines on FR4 substrate with the dielectric constant (*εr*) of 4.7 and thickness (H) of 0.8 mm, and loss tangent of 0.01. The ITS Butler Matrix microwave beamforming network was then extensively measured on a network analyser over a frequency range of 2.5 GHz to 3.5 GHz with the frequency of operation as 3.15 GHz. The Butler Matrix has been shielded with a metal box along with absorbing foam attached to the top lid of the inner box in order to reduce internal coupling and external interference effects. Fig. 10 indicates the fabricated ITS beamforming network being tested. Only one beam port and one output port are measured

0° Phase Shifter

8

5

6

7

**4.1 4×4 ITS Butler matrix microwave BFN realisation** 

45° Phase Shifter

Crossover

3

4

1

2

Hybrid Branch Line Coupler

Fig. 11 and Fig. 13 indicate the simulated and measured S-parameters exciting beam port 1 and beam port 2 respectively. At an operating frequency of 3.15 GHz, the simulated results agree with the measured results. As Fig. 12 and Fig. 14 indicate, the measured progressive phase shift for the Butler Matrix microwave beamforming network output ports exciting port 1 and port 2 respectively ensure the generation of four different beams at 3.15 GHz.

By using the narrow-beam signals available from the ITS Butler Matrix, it is possible for a vehicle to increase gain in the desired signal directions and reduce the gain in interference signal directions. The differences between the simulations and measurements and slight distortion of the beam shape might be due to the employed FR4 board and fabrication process, non-uniformity of matrix transmission line width, cross-coupling effects, and measurement errors. This Butler Matrix microwave network can be used as a planar passive BFN for multi-beam antennas used in automotive telematics and ITS applications.

The microwave beamforming network was designed to be placed anywhere on the envelope of the symmetrical cut-plane running through the centre of the vehicle. Possible antenna placement positions are therefore on the roof of the vehicle or inside the front and rear bumpers with a plastic radome. Antenna location is important to permit a mounting, which has little impact on vehicle styling, be of low cost, and be capable of addition to the vehicle with minimum re-design of surrounding components.

Fig. 10. Fabricated 4×4 ITS Butler Matrix Microwave BFN.

Fig. 11. 4×4 ITS Butler Matrix Microwave BFN Simulated and Measured S-Parameters Exciting Beam Port 1.

Microwave Beamforming Networks for Intelligent Transportation Systems 137

As Fig. 15 indicates, the proposed ITS Butler Matrix microwave beamforming network has been cascaded back-to-back in order to produce narrow-beams and broad-beams suitable for V2V and V2I automotive communications. Signals entering the input ports of the first Butler Matrix microwave beamforming network are subdivided into equal amplitude with progressive phase variation across the matrix output ports, for high-gain and narrow-beam

These signals are then fed into the Wilkinson power dividers. The signal from one end of each Wilkinson power divider forms a narrow output beam while the signals from the other ends of the Wilkinson power divider are fed into the second Butler Matrix network in order to regenerate the broad-beam signal characteristics of the individual radiating elements. As a result, high-gain and narrow-beam signals are produced on the output of the first Butler Matrix network while broad-beams are produced on the output ports of the second Butler

**4.2 Cascaded ITS Butler matrices microwave BFN design and performance** 

reception that are potential for ITS long-range automotive communication.

Matrix network that are suitable for ITS short-range automotive communication.

Fig. 15. Block Diagram of the Cascaded ITS Butler Matrices Microwave BFN.

second Butler Matrix microwave network.

network via an RF switch.

Fig. 16 and Fig. 17 present the computed array factor radiation patterns suitable for ITS long-range application and short-range application respectively. The concept of cascaded ITS Butler Matrices microwave beamforming network has been examined in which the first Butler Matrix microwave beamforming network will act as a power divider and the second Butler Matrix network will act as a combiner in order to produce high-gain narrow-beams for long-range communication from the outputs of the first Butler Matrix beamforming network and broad-beam signals for short-range communication from the outputs of the

Wilkinson

The ITS microwave BFN system can further be integrated with low-noise amplifiers (LNAs) to increase the gain and to reduce the noise power as well as an RF switch array for selection of the input beam ports of the network. A control circuit switches the RF switch to switch the oscillator signal rapidly among beam ports by changing the feed points at a specified rate. At the array ports, the phase shifted signals are amplified via an amplifier and then radiated through the antennas. On the receiving side, the receiver amplifies the signal and then a bank of filters filter the received signal which in turn is fed to the array ports of the

Fig. 12. 4×4 ITS Butler Matrix Microwave BFN Measured Phase Shift for Output Ports Exciting Beam Port 1.

Fig. 13. 4×4 ITS Butler Matrix Microwave BFN Simulated and Measured S-Parameters Exciting Beam Port 2.

Fig. 14. 4×4 ITS Butler Matrix Microwave BFN Measured Phase Shift for Output Ports Exciting Beam Port 2.

BM Port 1 Phase Output

<sup>2500</sup> <sup>2600</sup> <sup>2700</sup> <sup>2800</sup> <sup>2900</sup> <sup>3000</sup> <sup>3100</sup> <sup>3200</sup> <sup>3300</sup> <sup>3400</sup> <sup>3500</sup> �200

Fig. 12. 4×4 ITS Butler Matrix Microwave BFN Measured Phase Shift for Output Ports

Frequency (MHz)

BM Port 2 Insertion Loss

<sup>2500</sup> <sup>2600</sup> <sup>2700</sup> <sup>2800</sup> <sup>2900</sup> <sup>3000</sup> <sup>3100</sup> <sup>3200</sup> <sup>3300</sup> <sup>3400</sup> <sup>3500</sup> �20

<sup>2500</sup> <sup>2600</sup> <sup>2700</sup> <sup>2800</sup> <sup>2900</sup> <sup>3000</sup> <sup>3100</sup> <sup>3200</sup> <sup>3300</sup> <sup>3400</sup> <sup>3500</sup> �200

Fig. 14. 4×4 ITS Butler Matrix Microwave BFN Measured Phase Shift for Output Ports

Frequency (MHz)

 Measured S52 Measured S62 Measured S72 Measured S82

Fig. 13. 4×4 ITS Butler Matrix Microwave BFN Simulated and Measured S-Parameters

Frequency (MHz)

BM Port 2 Phase Output

 Measured S52 Measured S62 Measured S72 Measured S82 Simulated S52 Simulated S62 Simulated S72 Simulated S82

 Measured S51 Measured S61 Measured S71 Measured S81

> �18 �16 �14 �12 �10 �8 �6 �4 �2

Phase

dB Insetion Loss

Exciting Beam Port 1.

Exciting Beam Port 2.

Exciting Beam Port 2.

Phase

#### **4.2 Cascaded ITS Butler matrices microwave BFN design and performance**

As Fig. 15 indicates, the proposed ITS Butler Matrix microwave beamforming network has been cascaded back-to-back in order to produce narrow-beams and broad-beams suitable for V2V and V2I automotive communications. Signals entering the input ports of the first Butler Matrix microwave beamforming network are subdivided into equal amplitude with progressive phase variation across the matrix output ports, for high-gain and narrow-beam reception that are potential for ITS long-range automotive communication.

These signals are then fed into the Wilkinson power dividers. The signal from one end of each Wilkinson power divider forms a narrow output beam while the signals from the other ends of the Wilkinson power divider are fed into the second Butler Matrix network in order to regenerate the broad-beam signal characteristics of the individual radiating elements. As a result, high-gain and narrow-beam signals are produced on the output of the first Butler Matrix network while broad-beams are produced on the output ports of the second Butler Matrix network that are suitable for ITS short-range automotive communication.

Fig. 15. Block Diagram of the Cascaded ITS Butler Matrices Microwave BFN.

Fig. 16 and Fig. 17 present the computed array factor radiation patterns suitable for ITS long-range application and short-range application respectively. The concept of cascaded ITS Butler Matrices microwave beamforming network has been examined in which the first Butler Matrix microwave beamforming network will act as a power divider and the second Butler Matrix network will act as a combiner in order to produce high-gain narrow-beams for long-range communication from the outputs of the first Butler Matrix beamforming network and broad-beam signals for short-range communication from the outputs of the second Butler Matrix microwave network.

The ITS microwave BFN system can further be integrated with low-noise amplifiers (LNAs) to increase the gain and to reduce the noise power as well as an RF switch array for selection of the input beam ports of the network. A control circuit switches the RF switch to switch the oscillator signal rapidly among beam ports by changing the feed points at a specified rate. At the array ports, the phase shifted signals are amplified via an amplifier and then radiated through the antennas. On the receiving side, the receiver amplifies the signal and then a bank of filters filter the received signal which in turn is fed to the array ports of the network via an RF switch.

Microwave Beamforming Networks for Intelligent Transportation Systems 139

ITS is the application of high enabling technology to adaptive traffic signal systems control, congestion charging, information provision, and transit management systems in order to increase and enhance the safety and efficiency of the surface transportation system using radiowave beacons and real-time traffic information communication with major areas as: Multi-modal travel management and traveller information, commercial vehicle operations to achieve safe and cost-effective operation through cooperation and advanced automated

Vehicular telematics are a key technological component of future intelligent road networks. Such systems and technologies offer increased road efficiency, increased safety, improved communications and information services to drivers and passengers, and reduced road congestion and accident rates. Vehicle-to-infrastracture (V2I) or vehicle-to-vehicle (V2V) communication links are likely to be key elements (Fig. 18). Implementation of intelligent transportation systems and applications and vehicular telematics will require demonstration of a number of microwave systems and technologies at an acceptable price per unit. These technologies include antenna arrays, microwave beamforming networks, transmit/receive components, and a variety of sensors, both road and vehicle mounted, in order to increase road efficiency and provide additional services to drivers and travellers. The antennas and beamforming networks are required to provide steered and switched-beam smart radiation patterns to maintain links to moving vehicles and to compensate for signal fading in a complex and dynamic multipath environment. Hence, the development that will be a key to the provision of information-rich and high data-rate services will be microwave systems capable of providing communication links either with roadside beacons (V2I) or with other vehicles (V2V). In the latter case, it will be possible to form wireless vehicular ad-hoc networks (VANETs) with the benefit of reducing communication link range in high traffic

**5. ITS V2V and V2I automotive communications scenario** 

networking technologies, and advanced vehicle control and safety systems.

density and providing multiple routes between vehicles and roadside beacons.

Fig. 18. ITS Concept of V2V and V2I Automotive Communications.

Fig. 16. Cascaded Butler Matrices Computed Narrow-Beam Array Factor Radiation Patterns.

Fig. 17. Cascaded Butler Matrices Computed Broad-Beam Array Factor Radiation Patterns.

Fig. 16. Cascaded Butler Matrices Computed Narrow-Beam Array Factor Radiation Patterns.

Fig. 17. Cascaded Butler Matrices Computed Broad-Beam Array Factor Radiation Patterns.

## **5. ITS V2V and V2I automotive communications scenario**

ITS is the application of high enabling technology to adaptive traffic signal systems control, congestion charging, information provision, and transit management systems in order to increase and enhance the safety and efficiency of the surface transportation system using radiowave beacons and real-time traffic information communication with major areas as: Multi-modal travel management and traveller information, commercial vehicle operations to achieve safe and cost-effective operation through cooperation and advanced automated networking technologies, and advanced vehicle control and safety systems.

Vehicular telematics are a key technological component of future intelligent road networks. Such systems and technologies offer increased road efficiency, increased safety, improved communications and information services to drivers and passengers, and reduced road congestion and accident rates. Vehicle-to-infrastracture (V2I) or vehicle-to-vehicle (V2V) communication links are likely to be key elements (Fig. 18). Implementation of intelligent transportation systems and applications and vehicular telematics will require demonstration of a number of microwave systems and technologies at an acceptable price per unit. These technologies include antenna arrays, microwave beamforming networks, transmit/receive components, and a variety of sensors, both road and vehicle mounted, in order to increase road efficiency and provide additional services to drivers and travellers. The antennas and beamforming networks are required to provide steered and switched-beam smart radiation patterns to maintain links to moving vehicles and to compensate for signal fading in a complex and dynamic multipath environment. Hence, the development that will be a key to the provision of information-rich and high data-rate services will be microwave systems capable of providing communication links either with roadside beacons (V2I) or with other vehicles (V2V). In the latter case, it will be possible to form wireless vehicular ad-hoc networks (VANETs) with the benefit of reducing communication link range in high traffic density and providing multiple routes between vehicles and roadside beacons.

Fig. 18. ITS Concept of V2V and V2I Automotive Communications.

Microwave Beamforming Networks for Intelligent Transportation Systems 141

to form a state of the art vehicular network with enhanced performance to serve the ITS framework and objectives. The inherent capabilities of microwave beamforming networks and techniques together with VANETs complex algorithms and architectures will provide a powerful synergy for intelligent transportation systems and vehicular telematics realisation.

Fig. 19. ITS Vehicle Infrastructure Integration Architecture Equipment Types.

Fig. 20. ITS Wireless Access in Vehicular Environment (WAVE) System Components.

The need to relieve traffic congestion and make more efficient use of motorway networks requires a more sophisticated approach to traffic and transportation management. ITS applications and vehicular networks and telematics can offer many benefits using advanced RF and microwave technologies, where vehicles mounted systems communicate with other vehicles or with an infrastructure of roadside beacons. Hence, researches on intelligent transportation systems and applications were carried out to enhance safety and efficiency of road transportation related to vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) automotive communications. Microwave beamforming networks can greatly increase and enhance the performance of wireless systems used in intelligent transportation systems and framework. In this contribution, passive planar steerable microwave beamforming networks based on Rotman Lens and cascaded Butler Matrices have been designed and analysed in

**6. Conclusion** 

Intelligent transportation systems also play an important role in the research activities on road safety, allowing vehicles to detect a safety hazard and to react to it timely. Through immediate forwarding of hazard warning information to other vehicles via wireless vehicleto-vehicle (V2V) communication, other vehicles could avoid running into the hazardous situation. The same wireless communication interface could be used to provide the vehicle with traffic control and road safety information from roadside infrastructure via wireless vehicle-to-infrastructure (V2I) communication. Both V2V and V2I are the basis for ITS framework and applications providing a potential for avoidance of accidents.

In order to ensure efficient allocation of RF resources, it is important to group various V2Vand V2I-based applications in different categories based on their need for radio resources. The first such category is the ITS V2V-based Critical Road Safety Applications characterised by strict time constraints where one vehicle must warn another vehicle of a sudden safety hazard instantaneously. Such ITS applications have strict requirements on communication reliability, tolerable transmission latency, minimum throughput, and medium access delays.

Second category is the ITS V2I based Safety and Traffic Efficiency Applications which are informational applications. These applications may be less time-critical and may benefit from central resource management by roadside infrastructures, more RF link stability due to roadside unit's static nature, and better antennas. Depending on their unique requirements, Critical Road Safety Applications and Safety and Traffic Efficiency Applications require a higher QoS, such as instant access to RF frequency channel, high SNR and low channel interference, and reliable wireless communication to ensure that the safety messages are received by vehicles with high probability and for both types of applications, microwave beamforming networks based on Butler Matrices and Rotman Lenses can be employed in order to establish a reliable RF communication link among vehicular beacons.

A study was also carried out of Medium Access Control (MAC) protocols which are suitable for V2V and V2I automotive communications. The aim is to be able to communicate within a group of vehicles travelling as a cluster, between vehicles and a roadside transceiver, and from a roadside transceiver in a broadcast mode. Telematics architectures available in vehicular communication and networks are Vehicle Infrastructure Integration (VII) and Communication Access for Land Mobiles (CALM).

VII architecture seeks significant improvement in vehicle safety, mobility, and commerce by deploying a communication infrastructure on roadways and installing Dedicated Short Range Communication (DSRC) radios on all production light vehicles (Fig. 19). In this scenario, Onboard Unit (OBU) is located inside vehicle, Roadside Unit (RSU) is located on the road and acts as a data gathering and distribution point, control channel broadcasts application and vehicular communication establishment, and service control establishes communication between OBUs and RSUs and between OBUs. Also, DSRC which is a short to medium range communications service that supports both public safety and private operations in V2I and V2V communications is meant to provide very high data transfer rates for mobile wireless nodes in relatively small communication zones and with small latency.

Wireless Access in Vehicular Environment (WAVE) is the mode of operation used by IEEE 802.11 devices in the DSRC band allocated for ITS communication. Fig. 20 shows the WAVE system components. All these advanced wireless vehicular ad-hoc networks (VANETs) can further be integrated with advanced distributed microwave beamforming networks in order to form a state of the art vehicular network with enhanced performance to serve the ITS framework and objectives. The inherent capabilities of microwave beamforming networks and techniques together with VANETs complex algorithms and architectures will provide a powerful synergy for intelligent transportation systems and vehicular telematics realisation.

Fig. 19. ITS Vehicle Infrastructure Integration Architecture Equipment Types.

Fig. 20. ITS Wireless Access in Vehicular Environment (WAVE) System Components.

## **6. Conclusion**

140 Intelligent Transportation Systems

Intelligent transportation systems also play an important role in the research activities on road safety, allowing vehicles to detect a safety hazard and to react to it timely. Through immediate forwarding of hazard warning information to other vehicles via wireless vehicleto-vehicle (V2V) communication, other vehicles could avoid running into the hazardous situation. The same wireless communication interface could be used to provide the vehicle with traffic control and road safety information from roadside infrastructure via wireless vehicle-to-infrastructure (V2I) communication. Both V2V and V2I are the basis for ITS

In order to ensure efficient allocation of RF resources, it is important to group various V2Vand V2I-based applications in different categories based on their need for radio resources. The first such category is the ITS V2V-based Critical Road Safety Applications characterised by strict time constraints where one vehicle must warn another vehicle of a sudden safety hazard instantaneously. Such ITS applications have strict requirements on communication reliability, tolerable transmission latency, minimum throughput, and medium access delays. Second category is the ITS V2I based Safety and Traffic Efficiency Applications which are informational applications. These applications may be less time-critical and may benefit from central resource management by roadside infrastructures, more RF link stability due to roadside unit's static nature, and better antennas. Depending on their unique requirements, Critical Road Safety Applications and Safety and Traffic Efficiency Applications require a higher QoS, such as instant access to RF frequency channel, high SNR and low channel interference, and reliable wireless communication to ensure that the safety messages are received by vehicles with high probability and for both types of applications, microwave beamforming networks based on Butler Matrices and Rotman Lenses can be employed in

framework and applications providing a potential for avoidance of accidents.

order to establish a reliable RF communication link among vehicular beacons.

Communication Access for Land Mobiles (CALM).

A study was also carried out of Medium Access Control (MAC) protocols which are suitable for V2V and V2I automotive communications. The aim is to be able to communicate within a group of vehicles travelling as a cluster, between vehicles and a roadside transceiver, and from a roadside transceiver in a broadcast mode. Telematics architectures available in vehicular communication and networks are Vehicle Infrastructure Integration (VII) and

VII architecture seeks significant improvement in vehicle safety, mobility, and commerce by deploying a communication infrastructure on roadways and installing Dedicated Short Range Communication (DSRC) radios on all production light vehicles (Fig. 19). In this scenario, Onboard Unit (OBU) is located inside vehicle, Roadside Unit (RSU) is located on the road and acts as a data gathering and distribution point, control channel broadcasts application and vehicular communication establishment, and service control establishes communication between OBUs and RSUs and between OBUs. Also, DSRC which is a short to medium range communications service that supports both public safety and private operations in V2I and V2V communications is meant to provide very high data transfer rates for mobile wireless nodes in relatively small communication zones and with small latency. Wireless Access in Vehicular Environment (WAVE) is the mode of operation used by IEEE 802.11 devices in the DSRC band allocated for ITS communication. Fig. 20 shows the WAVE system components. All these advanced wireless vehicular ad-hoc networks (VANETs) can further be integrated with advanced distributed microwave beamforming networks in order

The need to relieve traffic congestion and make more efficient use of motorway networks requires a more sophisticated approach to traffic and transportation management. ITS applications and vehicular networks and telematics can offer many benefits using advanced RF and microwave technologies, where vehicles mounted systems communicate with other vehicles or with an infrastructure of roadside beacons. Hence, researches on intelligent transportation systems and applications were carried out to enhance safety and efficiency of road transportation related to vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) automotive communications. Microwave beamforming networks can greatly increase and enhance the performance of wireless systems used in intelligent transportation systems and framework. In this contribution, passive planar steerable microwave beamforming networks based on Rotman Lens and cascaded Butler Matrices have been designed and analysed in

**6** 

*Newcastle University United Kingdom* 

**Deploying Wireless Sensor Devices in** 

**Intelligent Transportation System Applications** 

A recent study by the UK Government's Office of Science and Innovation, which examined how future intelligent infrastructure would evolve to support transportation over the next 50 years looked at a range of new technologies, systems and services that may emerge over that period (UK DfT, 2006). One key class of technology that was identified as having a significant role in delivering future intelligence to the transport sector was wireless sensor networks and in particular the fusion of fixed and mobile networks to help deliver a safe, sustainable and robust future transportation system based on the better collection of data, its processing and dissemination and the intelligent use of the data in a fully connected environment. The important innovations in wireless and digital electronics are beginning to support many applications in the areas of safety, environmental and emissions control, driving assistance, diagnostics and maintenance in the transport domain. The last few years have seen the emergence of many new technologies that can potentially have major impacts

U.S. DOT recently launched a 5 Years ITS strategic research plan to explore the potentially transformative capabilities of wireless technology to make surface transportation safer, smarter and greener and ultimately enhance livability for Americans (US DOT, 2011). This research program formerly known as *IntelliDriveSM* and now renamed as "Connected Vehicle Research" program which focus to develop a networked environment supporting very high speed transactions among vehicles (V2V) and between vehicles and infrastructure components (V2I) or hand held devices (V2D) to enable numerous safety and mobility

The European Telecommunications Standards Institute (ETSI) has been creating and maintaining standards and specifications for Intelligent Transportation Systems (ITS) include telematics and all types of communications in vehicles, between vehicles (e.g. vehicle-to-vehicle), and between vehicles and fixed locations (e.g. vehicle-to-infrastructure). The ETSI not only looking for road transport domain but also the use of information and communication technologies (ICT) for rail, water and air transport, including navigation

As future intelligent infrastructure will bring together and connect individuals, vehicles and infrastructure through wireless communications, it is critical that robust communication

on Intelligent Transportation Systems (ITS) (Tully, 2006).

**1. Introduction** 

applications (US DOT, 2010).

systems (ETSI, 2009).

Kirusnapillai Selvarajah, Budiman Arief, Alan Tully and Phil Blythe

order to support the wireless systems used in vehicular networks, intelligent transportation systems, and collision avoidance program which includes rear-end collision avoidance system, intelligent adaptive cruise control, road departure collision avoidance system, and lane change collision avoidance system.

#### **7. References**

