Source traffic modeling in wireless sensor networks for target tracking

Researches around wireless sensor network (WSN) were very prolifc recently. However, traffic modeling related WSN research was poorly less. In this paper, source traffic dynamics in a simulated target tracking WSN scenario are explored. We find the source traffic arrival process doesn't follow the usually considered Poisson model. Instead, an ON/OFF model is found to be capable of capturing the burst nature of source traffic arrival. Further, we find the measured ON/OFF periods follow the generalized Pareto distribution perfectly. Mathematical analysis also shows a surprising fact: all ON/OFF period distributions in the experiment exhibit short-tail property, which is a nice property that could be exploited by applications such as anomaly detection and node failure detection.