Probabilistic Geographic Routing Protocol for Ad Hoc and Sensor Networks

In this paper, we present Probabilistic Geographic Routing (PGR), a novel approach for the problem of power-aware routing in wireless ad hoc and sensor networks. Our protocol uses only local information to probabilistically forward the packet to the next hop. Every node relies on a beaconing process to keep track of the changes in the set of its neighbors. In order to forward a packet, the node selects a subset of its neighbors. These candidate nodes are then assigned a probability proportional to their residual energy and the link reliability. We simulated PGR in NS-2 and compared the performance to two existing protocols, GPSR and Probabilistic Flooding. Based on the simulation results, PGR improves the throughput by 40%, increases the lifetime of the network by 30%, and decreases the overall end-to-end delay. In addition, we have implemented PGR on a real sensor network test-bed to verify our protocol.