**5.2. Rating approach**

278 Wireless Sensor Networks – Technology and Protocols

**5.1. Routing approach** 

security and energy.

**5. Comparison with previous work and main contributions** 

We provided a simulation based performance analysis for the efficiency of our proposed GETAR routing protocol. Simulation results proved the following points : GETAR improves the delivery ratio, decreases the drop and retransmission ratio and saves the retransmission power when compared with the previous work. The improvement in a performance metric can be achieved at different values of β parameter starting at a minimum value of β at a knee point in the curve. This value can be an optimum choice that guarantees best delivery ratio and better energy balancing. Energy balancing is negatively impacted by an increase in trust awareness. Thus, trade off considerations should be taken carefully in order to design an appropriate value of β. This will be subjected to the application preference between

In SAR[70], The routing operation needs to encounter a trusted route setup phase, which contributes some initial delay, especially with the crypto-based authentication required at the route setup. The trust metric used in SAR does not reflect nodes' behavior; rather, they represent a "rank" that a node exhibits based on its identity and various security service provision. Thus, a trusted node in SAR is a node that has the appropriate rank that meets the routing requirements. To rank a node is a problem by itself and requires crypto mechanisms. Our protocol, GETAR, is much simpler in that it assigns trust values to nodes based on nodes behavior. The routing decision rules in SAR are governed by the source, which makes the protocol less flexible. The routing decision is not to select the next hop but to decide to participate in the trusted route. As a result, selfish behavior is not addressed well in SAR. WSN constraints of power consumption are not treated. In fact, SAR targets ad

hoc networks with an assumption of more relaxed conditions as compared to WSN.

the routing operation itself since it relies on GPSR.

In TRANS[72], the trust, in fact, is associated with locations rather than nodes. The problem is that a location can be infected by a single node. The detour, then, will be around a larger area rather than a single node. "Innocent" nodes located in proximity of an infected location might be also isolated. If not, they are also exposed to heavy routing duties that may induce selfish behavior. TRANS is limited by single or multiple sink communication models. This assumption is necessary for the efficient operation of the protocol. Our proposed protocol, however, is more generic and can be applied to TRANS model or even for peer-to-peer model. TRANS discusses approaches to decrease energy consumption due to the security provision overhead. However, the protocol does not provide energy efficient techniques in

The RGR[73] protocol has no provision for energy efficiency as it relies on GPSR. The protocol totally relies on trust-based forwarding. If a node is completely surrounded by misbehaving nodes, there is no other mechanism proposed to select a next hop since all nodes will be eliminated from the node's forwarding list. RGR is a multi-path trust-based routing. Although multi-path is important for reliable services, we believe that it can be

energy consuming which we try to conserve in our work using GETAR.

The rating component of a reputation system deals with combining the first-hand and second-hand information meaningfully into a representative value. Moreover, it is responsible for updating such values as the behavior of nodes are evolving.

In literature, some rating approaches use a single value, called reputation, like CORE [5] and DRBTS [74]. This is similar to our approach in CRATER where we use a single value called the risk value, ri,j. Other rating systems like RFSN[2] and CONFIDANT[3] use two separate values, to represent the node reputation.

Some rating approaches updates the node reputation using both first-hand and secondhand information. In CRATER, we use this approach and we also introduce the neutral behavior period as another rating factor. Some other approaches like OCEAN (Observation-based Cooperation Enforcement in Ad Hoc Networks) [75] use just FHI.

In CRATER, SHI is accepted based on the cautious assumptions and the collected SHI by a node i about a node j is averaged to calculate a single ri,j,SHI. No validation check or honesty consideration is performed. However, some rating methods use validity and credibility tests for the gathered SHI. One method is to use a deviation test proposed in [64, 74].

Rating functions and mathematical modeling vary depending on the target applications. However, Beta distribution has been the most popular among researchers in reputation and

trust-based systems. It was first introduced in the field by Josang and Ismail [65]. Since then, many researchers have used the beta distribution including Ganeriwal and Srivastava [2] and Buchegger and Boudec [64]. In CRATER, however, we are using a simpler approach similar to the exponential average weighting. This is similar to the approach proposed in DRBTS [74].

When the weighing approach is used, an important issue in maintaining and updating reputation is how past and current information is weighted. For example, CORE tends to give more weight to the past observations assuming that a current observation should have a lower impact on a "greatly built history". On the other hand, RFSN tends to give more weight to recent observations based on the issue of aging. Aging means that we give higher weights to recent observations such that if you behave well you will survive more. As a result, malicious node will be enforced to reduce their attack to survive. In CRATER, we adopt the aging approach with some detailed modifications.

Up to our knowledge, there is no simple and global technique that can independently and efficiently evaluate reputation systems or rating components in the context of WSN and ad hoc networks as compared to our proposed technique, RESISTOR. However, the work in [61] proposes an attempt on comparing reputation systems quantitatively based on game theory. The authors, thus, identify different notions of reputation systems like, contextualization, personalization, individual and group reputation, and, direct and indirect reputation. But, it is more complicated than RESISTOR. Moreover, RESISTOR can be used as an indicator to understand the flaws or plus points in the rating system.
