부산대학교 도서관

Several problems are required to be fixed in order to apply geographical routing protocol greedy perimeter stateless routing (GPSR) in wireless sensor networks. First, GPSR is designed under the assumption of symmetric links (i.e., bidirectional reachability), which is not realistic for many practical sensor networks, since wireless links in sensor networks often are asymmetric. Second, in sensor networks, packet destinations are often marked with locations instead of identifiers like IP address. Therefore, packets are routed to the home node, which is the node geographically closest to the destined location. Under GPSR, when the target location is outside the exterior perimeter of the sensor network, each packet has to visit all nodes on the boundary of the sensor network topology before it identifies the home node, which results in energy inefficiency. Third, due to the dynamic nature of sensor networks, maintaining data consistency, that is, data retrieved from the home node for a location should be consistent with the data sent to the same location, becomes a challenge when home nodes change. We propose on-demand GPSR (OD-GPSR), a data driven geographical routing protocol customized for sensor networks with solutions to the above three problems. Simulation results show that OD-GPSR performs well in terms of energy efficiency and packet delivery rate at the cost of a little bit more packet delivery delay.