Having one's cake and eating it too: Better routes and lower control traffic for AODV

In this research we develop new techniques for optimizing the performance of a reactive routing protocol in operational environments characterized by high node mobility and long-lived connections. The question we seek to answer is whether in such environments, reactive routing protocols necessarily exhibit a tradeoff between control traffic and route optimality. More specifically, does a protocol which makes use of less control traffic (i.e. better) than standard AODV, necessarily exhibit connection routes that are longer (i.e. worse) than those achieved by standard AODV? We show that the commonly assumed tradeoff can be avoided, and that it is possible to "Have one's cake and eat it too". Towards this, we design an extension of the AODV protocol, and show through extensive ns2 simulation experiments that the new protocol both significantly reduces the control traffic overhead, while simultaneously improving the topological optimality of connections. These remarkable conclusions are seen to continue to hold scalably as one varies situational parameters such as network size, number of connections, and node mobility.