| A mobile ad hoc network is an autonomous system of mobile hosts connected by wireless links. Such networks are useful in military and other tactical applications, e.g., emergency rescue or exploration missions, where cellular infrastructure is unavailable or unusable. Because of constantly changing topology, ad hoc networks must adopt efficient dynamic routing techniques. Routing overhead is one important concern as the wireless link bandwidth is typically low. Recently, a new class of protocols called “on-demand” protocols has received attention because of their potential for low routing overhead. On-demand protocols discover routes on an “as-needed” basis, rather than in a more proactive fashion, as in the traditional link state or distance vector protocols. This dissertation presents a comparative simulation study of on-demand protocols with the more traditional proactive protocols on a common platform across a range of traffic and mobility scenarios. The simulation results show that on-demand protocols indeed demonstrate low routing overheads, but over-reliance on query flooding can actually cause on-demand protocols to lose most of this overhead advantage at high loads.; As a countermeasure, we develop and evaluate two techniques that reduces the routing overhead for on-demand protocols. The first technique, called query localization, uses certain locality heuristics to prevent network-wide query flood. It effectively limits the query to a small region where the route is very likely to be found. The second technique, called multipath routing, focuses on reducing the frequency of query flooding by exploring multiple, disjoint routes per flood operation. This provides the routing protocol with alternate routes when the primary route breaks. Simulation results show that either of these techniques can reduce routing overheads significantly.; Unlike stationary hosts, mobile hosts can operate only as long as the their batteries last. Thus, battery power conservation is an important component of a routing protocol. We describe a “power-aware” mechanism that can work with any on-demand routing protocol to offload the routing responsibilities from the nodes running low on battery power towards the nodes with more power. The simulation results demonstrate that this technique distributes the available battery power more evenly and improves the operation lifetime of the ad hoc network. |
