Loss Temporal Dependency Tomography in Wireless Sensor Network

Due to the inherent stringent bandwidth and energy constraints, it is usually impractical to directly collect link loss statistical data from each node in sensor network. Here we consider the problem of inferring the internal link loss characteristics from end-to-end measurement. We use the Gilbert error model to model the sensor link losses, formulate the problem of link loss characteristics estimation as a Bayesian inference problem, and propose a MCMC algorithm to solve it. The simulation shows that the link loss performance parameters can be inferred accurately, and the proposed algorithm scales well according to the sensor network size.