On the route refresh frequency for on-demand battery life routing in ad hoc networks

Loss of connectivity to even node is significant in peer-to-peer ad hoc networks as any node can become a source or destination. Accordingly, we define the network lifetime as the time by which the first node runs out of battery power. Conditional min-max battery cost routing (CMMBCR) [C.K. Toh, 2001] is a generic power-aware routing algorithm proposed to strike a balance between the contrasting objectives of maximizing the node lifetimes and minimizing the energy consumption. CMMBCR works by defining a battery protection threshold γ to switch between a routing scheme in which the total energy consumption is minimized and a routing scheme in which the minimum residual energy of any node on the route is maximized. In an on-demand power-aware routing protocol, existing routes have to be changed/refreshed to take into account of the available battery power of the nodes and extend the time at which the nodes run out of their battery power. Very few works have studied the performance of an on-demand distributed version of CMMBCR. We claim that the route refresh frequency is critical to the performance of CMMBCR and it has to be chosen depending on the battery protection threshold γ and the level of node mobility. We also claim that when the network topology changes dynamically (moderate to high mobility), there is no need to do route discovery to refresh an existing route. Mobility itself guarantees that an existing route is short-lived and the residual battery levels of the nodes can be learnt when a route discovery is done due to route failures. Our simulation results are based on implementing CMMBCR on the top of DSR in ns-2 [K, Fall et al, 2001]. We study the impact of the route refresh frequency on the network lifetime and end-to-end delay per packet under different values of γ, node mobility and offered traffic load. The results presented in this paper can aid in the on-demand performance study of existing power aware routing algorithms in mobile ad hoc networks.