Fault detection in mobile ad-hoc networks

Abstract

In this thesis, we consider the problem of self-diagnosis of mobile ad-hoc networks (MANETs) using the comparison approach. In this approach, a MANET consists of a collection of n independent heterogeneous mobile hosts interconnected via wireless links, and it is assumed that at most sigma of these mobile hosts are faulty. In order to diagnose the state of the MANET, tasks are assigned to pairs of mobiles and the outcomes are compared. The agreements and disagreements among mobiles are the basis for identifying the faulty ones. The comparison approach is viewed as one of the most practical diagnosis approaches. We have developed two distributed self-diagnosing protocols (DSDPs) for MANETs the first (Dynamic-DSDP) is more energy efficient, since its communication complexity is lower, while the second (Adaptive-DSDP) has a lower diagnosis latency and is more appropriate for MANETs that have more dynamic topologies. Correctness and complexity proofs are provided. Using the ns-2 simulator, we implemented the two protocols, which provided us with further insight into their effectiveness and illustrated the effect the number of faults had on their efficiency.