Detour Routing Protocol for Geographic Sensor Networks

Abstract

Routing in sensor networks is a challenging research topic due to the barriers (holes) on the natural topography often causing routing algorithms to fail. Most geographic routing algorithms in sensor networks apply the greedy forwarding method to discovery a path to the destination without the global states. However, greedy forwarding could fail because the local minima problem. In this paper, we propose a novel algorithm called Detour Routing based on Quadrant Classification (DRQC), to reduce the occurrence of the local minima problem by avoiding sending packages to the nodes which could have the local minima problem. The basic idea of DRQC is based on that each node knows the geographic information of its 2-hop neighbors, and then each node decides its state: red or white. A red node has no local minima problem, but a white node could have local minima problem. DRQC requires that each node chooses one of its 2-hop neighbors, which is red and whose distance to the destination is the shortest one, as the next hop for forwarding a packet. If no such a neighbor exists, then traditional 2-hop Greedy scheme is applied to select the next forwarding node. We implement the DRQC algorithm and compare it with two algorithms: 1-hop greedy and 2-hop greedy. Simulation results show that our algorithm (DRQC) has higher packet delivery rate (96%) than the two algorithms: 1-hop Greedy (79%) and 2-hop Greedy (87%).