**2. Optical fiber sensing**

Light travels within the optical fiber and constrains itself in the medium due to the total internal reflection mechanism of light signals. This total internal reflection is

possible if the light tries to pass from the highly dense medium to a lower one and the concept of this principle is depicted in **Figure 1**. **Figure 1** shows that a deflection away from the reference occurs if the light signal passes from a highly dense medium to a low dense medium. If the incident signal surpasses the critical angle (*θ*<sup>3</sup> in our case), the total internal reflection occurs. The same figure shows the total internal reflection occurs at the incident angle *θ*5. Within an optical fiber, the light traveling follows the principle of total internal reflection. The two main components of an optical fiber are a core and a cladding, as depicted in **Figure 2**. The core is the highly dense medium, whereas the cladding is a low dense medium. Light is injected from one end of the optical fiber and due to total internal reflection, the light retains within the core until it reaches the end of the optical fiber.

The conventional vehicle and track monitoring systems are mostly based on electrical principles. The main use of optical fibers is in the telecommunication industry. However, the idea of OFS in railway systems was engendered to avoid EMI, which is considered to be a more challenging impairment than any other drawback of electrical-based systems. Later on, many other OFS-based solutions were used in comparison to the conventional sensing techniques in the railway industry. By that time, the OFS solution got popular, and many challenging issues of the railway infrastructure were resolved with the OFS solutions, especially in the last two decades.

DOFS is a specialized group of technologies under OFS, that provide sensing at any location of the fiber, where needed and the sensing can be done simultaneously at

**Figure 1.** *A demonstration of total internal reflection.*

**Figure 2.** *Cross-section of an optical fiber cable.*

#### **Figure 3.**

*Typical operating spectrum of different distributed optical fiber sensing technologies.*

multiple locations. The DOFS-based technologies are grouped based on three types of light wave scattering. These are Rayleigh scattering, Brillouin scattering, and Raman scattering [1]. The frequency band of each of these scattering types is different, and an overview of these bands is depicted in **Figure 3**.

All of these scattering types are comprised of many other sensing technologies. For example, the technologies associated with the Rayleigh scattering include distributed acoustic sensing (DAS) or phase optical time-domain reflectometry (*ϕ*-OTDR), and polarization-OTDR. In regard to the railway industry in the last decade, the focus in the railway sector is mainly on DAS as compared to other Rayleigh-based scattering technologies.

DAS can be used in distributed and quasi-distributed nature sensing, and there is a possibility of dynamic stress sensing for both the railway vehicles and the infrastructure. Moreover, due to the possibility of supporting a very long range, train tracking, and operational monitoring have achieved a high level of maturity in the research field associated with DAS technology for its use within the railway sector. On the other hand, Brillouin scattering-based DOFS technologies include Brillouin optical time domain reflectometry (BOTDR), Brillouin optical time domain analysis (BOTDA), and Brillouin optical frequency domain analysis (BOFDA). However, due to the hybrid sensing capability of Brillouin scattering-based technologies can be rarely used in railway infrastructure monitoring as the high temperature of the railway tracks may affect the measured stresses. Beside, Raman scattering-based sensing technologies are not very popular in the railway industry due to their only sensing capability for temperature and relatively less response to a rapid change in a measured quantity. Moreover, Raman scattering-based sensing technologies only measure the temperature in a distributed manner, but this technology can be expensive for use in the railway sector. Many other OFS-based technologies, such as interferometry-based sensing techniques, can be regarded as unreliable in the railway industry due to their nonlinear nature, and hence cannot be used to meet the required objectives. One of the advanced interferometry-based sensing technologies is the optical frequency domain reflectometer (OFDR), which is not popular in the railway sector due to its short-distance applications. It is due to these reasons that the use of FBGs and DAS technologies is dominant in the railway sector, and more. than 95% of the OFS-based

research articles are based on these two technologies. Therefore, this chapter provides insight into both the FBGs and DAS technologies for their extensive use within the railway industry.

### **2.1 Fiber Bragg gratings in railways**

FBGs are the best quasi-distributed alternative in the OFS field, and their use has become quite popular within the railway industry in the last two decades. As the FBGs can support a very long range of distances, and hence this technology has attracted the researchers' attention in the railway industry. Moreover, the use of any FBG type can provide the dynamic sensing capability, which has attracted its interest in the railway industry.

With distributed sensing, the whole optical fiber can act like an array of continuous sensors at virtually no gap among the sensing points. Moreover, the same fiber can be used as a communication medium, and hence the ease of installation make these sensors a favorite choice. On the other hand, the quasi-distributed sensing points at a relatively low gap among the sensing points can be achieved with ease of installation with the help of FBG. These sensors are normally the best alternative to the conventional sensing used within the railway sector. FBGs are formed with the help of an intense optical interference pattern within a fiber core. These patterns are formed such that the grating acts like a periodic perturbation of the refractive index. The gratings can perform many functions such as filtering, diffraction, and reflection. However, the most important property of these sensors is the reflection of incident light waves according to a predefined wavelength. The FBGs can be embedded within the optical fiber at discrete positions, as shown in **Figure 4**. These sensing points are generally comprised of a periodic modulation of the refractive index, which is normally embedded in the core of a single-mode optical fiber. There are two types of these gratings, uniform and nonuniform grating. The phase fronts of the uniform grating are usually vertical to the longitudinal axis of the fiber. After the light strikes along the grating plane, it gets scattered along the core part of the fiber, as shown in **Figure 4**. The period of modulation index plays a very important role in controlling

**Figure 4.** *A demonstration of FBG, passing and reflecting the selective frequency components.*

the width of the frequencies scattered by the grating, and this index is represented by Λ, which can be defined as

$$
\Lambda = \frac{\lambda\_B}{2n\_{\text{eff}}} \tag{1}
$$

Where *λ*<sup>B</sup> is the Bragg wavelength of the input light, which is back-reflected from the grating plane, and *n*eff is the effective index of the optical fiber.

The important aspect of the grating period is that the wavelength of the reflected light is modified in accordance with the grating period due to a change in external environmental effects such as temperature or applied stress. It is due to this reason that a change in the external environmental conditions is proportional to a change in the refractive index of the core and the modulation index Λ, and this, in turn, provides a proportional change in temperature and applied stress. In other words, a change in both the external variables (stress and temperature) brings an offset to the Bragg wavelength *λB*, and the whole process is demonstrated in **Figure 4**. This figure depicts that the light waves are reflected according to the predefined wavelength gratings, and hence pass the remaining wavelengths of light. The reflected wavelength is processed such that the environmental effects such as strain or temperature are numerically acquired, and hence detection of a change in wavelength provides useful information about the measured variable.

The use of FBGs is becoming quite popular in the railway industry in regard to the monitoring and inspection of both the vehicle and track and the previous research shows a great deal of work in this area than DOFS as far as the railway sector is concerned. FBGs are popular in railway operational monitoring due to their small weight and volume, ease of handling, high spatial resolution, high precision and accuracy in the numerical results, and the capability to multiplex multiple signals simultaneously due to the wave division multiplexing capability of the fiber optics. By using a spacing between each FBG sensing point of less than 1 km, it is possible to extend the measurement range up to 100 km.

Literature reveals that FBGs in the railway sector have been used for the first time in the year 2004 for derailment detection [2] and axle counting [3]. Since then, these sensors have been used in a myriad of different applications in the railway infrastructure and the associated vehicles. The derailment detection was used as a metric for identifying different train types, whereas axle counting was used as a metric to detect the vehicle speed. After the initial steps were taken, the railway sector used FBGs in a number of applications such as the detection of defective wheels [4, 5]. These sensors were installed in the vicinity of the sleepers for health monitoring of both the rails and the vehicle's wheels status and proved the metrics such as "infrastructure measuring infrastructure metric" and "infrastructure measuring vehicle metric." The two metrics measurements were possible with the help of elegant signal processing techniques, which was otherwise impossible without the use of FBGs. Moreover, the train moving direction and axle load were measured along with many other important parameters in [6]. Beside the rail and infrastructure monitoring, the FBGs were utilized by bonding them to the railway tracks in order to measure speed and track the moving vehicles along the whole distance [6–8]. Beside, the innovative FBG interrogator has been designed to measure the train speed and axle load with a minimum possible number of FBG sensors [6]. Due to the ease of installation of FBG sensors and systems, the applications of these sensors in the railway sector did not end till the stage of limited monitoring as we can see in the case of conventional sensing systems. These sensors can be installed in brack blocks, wheels, axles, and bogies in order to complete a composite sensing system, as installed by Mi in [9]. Such a composite sensing system can provide a complete monitoring system, which was otherwise impossible without the use of FBGs. Aside from these, different force types such as longitudinal and vertical forces on the railway tracks are possible with FBG [10, 11]. To filter out the impact of temperature from the strain measurements, efforts were made in [12], and hence there is a possibility of a provision of pure strain measurements even in the harsh environment where a large variation in temperature is possible. The benefits of FBGs in the railway industry are not limited to the mentioned applications. Vibration measurement in geogrid-reinforced ballast and unreinforced ballast along with the lateral displacement were made [13]. Moreover, FBGs can be utilized for the differential settlement of railway tracks [14]. Additionally, the operation monitoring/inspection of switchblades, fishplates, and stretcher bars is possible with the help of FBGs [15]. An important part of the railway infrastructure is the railway bridges. The effects of transverse vibration, dynamic load bending, and vertical deflection of the railway bridges were inspected in [16]. FBGs were proven to replace the electrical based sensing systems such as strain gauges, and this was verified by inspecting the vertical acceleration and contact force in pantograph catenary [17]. **Table 1** refers to different railway-specific applications associated with the use of FBGs.



#### **Table 1.**

*Applications of FBGs in references to previous work.*

#### **2.2 Distributed acoustic sensing in railways**

DAS works under the principle of *ϕ*-OTDR, and it is a Rayleigh scattering-based DOFS technology used to sense the vibrations and perturbations at a regular or selected spatial point along the entire length of the optical fiber. Each sensing point is analogous to thousands of hydrophones connected in series. According to the working principle of DAS, the light signals from a laser source (around 1500 nm wavelength) are modulated with rectangular pulses, and these pulses are sent to the fiber under test (FUT). Each segment of the FUT reflects the Rayleigh backscattered signals, and a circulator is used to divert these back-reflected signals to the direction other than the one from which these pulses originate. At the intended port of the circulator, the photodetector is installed, which converts the modulated pulses into electrical signals. A positive aspect of the DAS system is that the simultaneous perturbed signals can be retrieved without any specialized signal processing system. The two most commonly used configurations of a DAS system are directed detected and coherent detected systems, as shown in **Figures 5** and **6** respectively. The configuration in **Figure 5** is the direct detected system that implies there is no reference light source at the photodetection stage. A coherent detected *ϕ*-OTDR system is one in which the laser source acts as the reference signal to provide additional phase information at the photodetection stage. Converting the direct detected system to a coherent detected system requires a modification such that the laser source is divided into two parts with the help of a coupler. One of the branches of this signal is injected into the optical modulator, whereas another part of the same signal leads to the balanced photodetector, as depicted in **Figure 6**.

In both the direct and coherent detected systems, the optical modulator is considered to be any type of acousto-optic modulator (AOM) or electro-optic modulator (EOM) with a high extinction ratio (a ratio between transmitted one and transmitted zero). The modulation drops the signal strength many folds, and therefore an Erbium-

#### **Figure 5.**

*A direct detected ϕ-OTDR system.*

**Figure 6.**

*A coherent detected ϕ-OTDR system.*

doped fiber amplifier (EDFA) is used to amplify these signals. The EDFA and filters are optional and these devices are installed if needed in the very large *LFUT* (length of FUT) applications. Each injected pulse with a predefined pulse width (PW) within the fiber defines the spatial resolution (SR) of the measured distance within which the perturbation is felt. These pulses are injected within the fiber with a specific pulse repetition rate (*fPRR*), and the backscattered signals are acquired with the sampling frequency of *fDAQ*. The maximum frequency (*fPER*) from the frequency components generated due to applied perturbation is another factor, that is. restricted due to the associated parameters. Here is a list of compromised parameters that depend on another parameter and are mentioned below with examples (**Table 2**).

Two types of sampling rates are defined in the abovementioned discussion. One of these types relates to the spatial sampling rate, and its speed depends on the sampling frequency of the data acquisition (DAQ) card, termed *fDAQ*. The second type relates to the temporal sampling rate, and its speed depends on PRR. From the length of a fiber, one can decide the maximum allowable frequency of the perturbation. The smaller the length of the fiber, the higher will be the frequency at which the perturbation can be


#### **Table 2.**

*Threshold of the limiting parameters.*

detected. Both the SR and the DAQ sampling rates are required to define the minimum spacing between two sensing points. The lower the SR and DAQ sampling rate, the smaller the spacing between two sensing points. Moreover, the lower the SR and DAQ speed, the smaller the spacing between two sensing points. After sending a single pulse, the fiber provides a response signal and each of these samples correspond to a distance 1/PRR in the time domain, where the mentioned response signal is the result of Rayleigh backscattered signals along the whole FUT. The correspondence distance of this response in meters can be obtained using the formula: *LFUT* = *c* 2 *neff fPRR*. The response against the whole FUT is called a single trace, which is a stationary random process, and this fact can be observed with the help of observing each received trace provided that the fiber is not disturbed with the assumption of the very long linewidth of the laser source. Differentiating one received from its previous counterpart is termed the differential data-trace. One can easily observe the phase change due to the applied perturbation with the help of a differential data-trace. In case of any perturbation, multiple traces are normally acquired, and the difference of each subsequent trace is taken in order to determine the point of perturbation.

**Figure 7.** *Perturbation demonstration using differential signals of ϕ-OTDR system.*

A schematic diagram showing the behavior of differential signals due to a stretch of the fiber by an applied perturbation is shown in **Figure 7**.

DAS is a revolutionary photonic sensing technology that exploits the use of a standard communications fiber into a linear array of discrete vibration sensors. Activities such as people walking/running, hot-tapping pipelines, pipeline leakage detection, perimeter intrusions, moving vehicles, industrial operations, failing mechanical components, firing direction detection, and many applications are responsible for generating vibrations with distinct acoustic characteristics. DAS technology monitors these vibrations and accurately detects, classifies, and reports on the vibration events. With DAS, there is no need to install the conventional sensors, and a simple G.652 type single-core fiber optic cable is enough to sense the whole distance in a distributed fashion, and hence it saves a huge amount of cost among hundreds of kilometer distance spans. The efficient algorithms requiring very few data traces [31–33] can help to calculate the instantaneous speed of the trains. As we know, terrorist acts along railway tracks, derailments, and train collision accidents were quite common in the history of railway-based accidents; therefore, DAS can play a very important role to avoid these accidents in the future. The use of DAS technology in the railway sector is not only a cost-effective solution but due to the provision of enormous data at each spatial location also there is a possibility of involving artificial intelligence to automate both the security and train tracing along the whole railway track, which is normally in hundreds of kilometers range. Fiber response or traces received against a few injected pulses are mostly insufficient to detect a certain

### *Distributed Optical Fiber Sensing in Railway Engineering DOI: http://dx.doi.org/10.5772/intechopen.111564*

perturbation as most differential data traces do not describe the effect of external perturbations. Most often, the machine learning algorithms applied to these differential data traces do not provide essential information for classifying different types of intrusions unless a myriad number of data traces are acquired. Normally, a single data trace is received with a certain delay after the injection of a single pulse within an optical fiber. A large percentage of the received data traces are irrelevant, and these irrelevant data traces are also received with the same delay. Hence, DAS can be successful in many applications for which the delay is not important. However, in the case of high-speed moving trains, where instantaneous intrusion detection, instantaneous speed determination of the trains, and the instantaneous location of each wheel of the respective bogies of the whole train are solicited, the lag in the DAS system is unacceptable. There are many applications of the DAS system, including fence, border, and pipeline security systems. Several signal processing methods applied directly on differential data traces include time-series-based algorithms [34–38], and frequency-based algorithms [39–41] approaches were suggested to provide a better probability of detection and classification accuracy in perturbation detection and event recognition applications respectively without imparting emphasize on utilizing a smaller number of data-traces efficiently. The time-frequency-based approaches such as discrete wavelet transform [42–44], Hilbert-Huan transforms [45], or similar algorithms [46–48] are best suited for trace-to-trace fluctuation-based noise alleviation [49, 50] than the frequency-based or time-series-based approaches. A drawback of the time-frequency-based approach is that these techniques are intensively parameter-dependent. For example, choosing a very selective mother wavelet and a vanishing moment in DWT-based algorithms for each specific event in a perturbation recognition application may not be possible, although, these algorithms can be suitable for the applications such as perturbation detection. The issues relevant to sampling data relevant to high-speed vehicles were suggested for the first time in [31–33], and hence now it is possible to consider DAS for very high-speed vehicles.

A detailed insight into DAS technology with respect to the railway infrastructure and the associated railway-based vehicles, whether trains or trams, is discussed here. DAS was first introduced by Juarez in the year 2005 while demonstrating the concept of intrusion detection, and this class of distributed sensing became popular afterward. Initial work was carried out on the ballastless track structure in the year 2013 [51] followed by speed and position detection in the year 2014 [36]. As mentioned before, the two main applications of any railway system are train traffic management and operational monitoring. The research work in the case of train traffic management includes speed and position precision improvement along with external applications such as security and other activity monitoring alongside railway tracks, whereas the operational monitoring applications involve continuous monitoring of both the train and tracks for their faults. **Table 3** includes the research work by exploiting DAS and *ϕ*-OTDR in traffic management of the railway system. Another important set of railway applications involves the operational monitoring of both the trains and tracks. **Table 4** shows the literature work regarding the use of the DAS system for operation monitoring of the railway system.
