"Energy Efficient Data Collection in Sensor Networks with Multiple Mobile Access Points"
P. Venkitasubramaniam and L. Tong
39th Annual Conference on Information Systems and Sciences, Princeton, NJ, March 2005.

We consider the problem of data col lection from sensor networks using multiple access points. These access points could be intelligent and powerful sensors acting as cluster heads or just fixed collecting stations covering the network. The cover- age areas of the access points may overlap, making transmissions from overlapping areas more energy efficient. With energy efficiency as the metric, we study cross-layer strategies where a fraction of the data is routed across regions to areas covered by multiple access points and the remaining are directly transmitted to the APs. Using ALOHA as the random access protocol, we show that for any given routing protocol, the optimal energy efficient strategy undergoes a phase transition with respect to net traffic load: at low traffic load, the most energy efficient strategy is to route all packets to the overlapping region, whereas, at high traffic load, routing does not reduce energy consumption. Furthermore. the phase transition point is shown to be independent of the fraction of overlap.