Using Flow Cost for MANET Routing

Abstract

This thesis has introduced the concept of Flow Cost (FC) that can capture many metrics of interest, but not just the shortest distance or smallest delay as done in many traditional algorithms. It allows us to capture the influence of flows from other parts of the network through their interaction at a given node. We first proposed the routing algorithm called GANFA in static mobile ad hoc networks where FC is formulated as a function of node power, link delay and hop count (and later, queueing delay). The FC also allows us to choose the best path through optimization, and to provide bandwidth allocation. We have further extended our study to the dynamic mobile ad hoc network by proposing the FC-AOMDV multipath routing algorithm which prepares a back-up alternate path at the same time when the primary route is set up. A comparison of video application between the FC-AOMDV and the AODV is done to prove the performance improvement. With the consideration of real-time transmission, a predictive real-time approach named RPET-FC-AOMDV is designed that studied the influence of both the node speed and node density. Besides applying FC in 2D network routing, we have also used it to design the FC-3DGR for AANET application in which we have explored the localization and high-speed issues in 3D MAENTs. Unlike using distance as the only criterion in greedy protocol, our FC accounts for both distance and number of neighbors for routing path selection. In this way, we can avoid some problems such as VNP (Void Node Problem) encountered in other greedy routing study.