On-Demand Link-State Routing in Ad-Hoc Networks

This thesis explores the challenges, merits and demerits of using link-state information for on-demand routing in ad hoc networks, such that routers maintain path information for only those destinations for which they have data traffic. We first present the source tree on-demand adaptive routing SOAR protocol, in which each router exchanges with its neighbors a source tree containing paths to only those destinations for which the router is the source or relay of data packets. The main advantage of SOAR is that it is more scalable and better performing than current state-of-the-art on-demand routing protocols. However, a limitation of SOAR is that it requires data packets to specify the paths they traverse to detect loops. To eliminate the need for source routing or path traversal information in data packets, we introduce the on-demand link-vector OLIVE protocol, which prevents temporary loops for each destination by synchronizing relevant link-state information among neighbors. In OLIVE, the advertised paths combine to form a source graph, rather than a source tree. OLIVE is shown to outperform the current routing protocols proposed for mobile ad-hoc networks in terms of control overhead, throughput and network delay.