Effect of limited topology knowledge on opportunistic forwarding in ad hoc wireless networks

Opportunistic forwarding is a simple scheme for packet routing in ad hoc wireless networks such as duty cycling sensor networks in which reducing energy consumption is a principal goal. While it is simple and can be analytically characterized, it suffers from a high end-to-end latency. In this paper we show how this latency can be drastically reduced if nodes have limited knowledge of network topology (that can be achieved by scoped dissemination of link state information), and hence deriving a hybrid routing protocol. We give an analytical formulation of end-to-end latency between any pair of nodes in such duty cycling networks as the scope of topology dissemination is varied. We borrow from our prior results derived from spectral graph theory to derive exact expressions for mean latency as a function of various network and protocol parameters such as size, duty cycle probability, and scope of link state dissemination. These analytical expressions agree very well with simulation results. We also show how this latency analysis can be coupled with overhead analysis to determine good values of topology dissemination scope.