Analysing the performance of MANET trust-based routing protocols in the presence of attacks

Abstract

Mobile ad-hoc networks (MANETs) are wireless multi-hop networks that do not rely on any fixed infrastructure, unlike traditional networks. Nodes in MANETs are formed dynamically and are free to move in any direction at variable speeds. MANETs can be used efficiently in situations in which infrastructure has been destroyed, such as in earthquakes, or where there is no fixed infrastructure, such as in wild areas, or where the existing infrastructure is insufficient, e.g. for a festival or other event. MANETs do not have a centralised administration and nodes join and leave the network freely without any validation or authentication. These special characteristics make MANETs vulnerable to many types of network attack that can negatively affect their performance. Moreover, because of their nature, it is not possible to employ the solutions designed for traditional networks. One possible solution that could help assure the performance of MANETs in the face of a network attack is to implement trust management. The principle of trust management in MANETs is that each node monitors the behaviour of its neighbouring nodes and tries to detect any malicious activities. Once a node identifies that a neighbouring node is malicious, it will categorise it as untrustworthy and avoid sending data to it in the future. There is more than one way of designing trust management schemes in MANET. One is a direct trust scheme, in which each node calculates the trust values for each of its neighbouring nodes itself. Another is an indirect trust scheme, in which a node receives recommendations about its neighbouring node from other nodes in the network. This study proposed and implemented four direct trust schemes and evaluated their performance against black-hole, grey-hole, selfish, and flooding attacks. It also proposed and evaluated the performance of an indirect trust scheme against a black-hole attack. The NS2 and NS3 network simulators were used to run the simulations. The mechanisms were implemented with the most common MANET routing protocol, ad-hoc on-demand distance vector (AODV). Having added the trust management schemes to the AODV protocol, the study compared the performance of the plain AODV and the trust-based AODV in the presence of the attacks. The simulations showed that implementing a direct trust management mechanism within the AODV protocol improved the performance of MANETs in the presence of attacks at the cost of a slight increase in the overhead. The indirect trust scheme resulted in greater improvements in performance than the direct trust schemes but with higher overhead.