Distributed sequential Monte Carlo algorithms for node localization and target tracking in wireless sensor networks

We address the problem of tracking a maneuvering target that moves along a region monitored by a wireless sensor network (WSN) whose nodes, including sensors and data fusion centers (DFCs), are located at unknown positions. Therefore, the target trajectory, its velocity and all node locations must be estimated jointly, without assuming the availability of any “beacons” with known location that can be used as a reference. We introduce a new method that comprises: (i) a combination of Monte Carlo optimization and iterated importance sampling to yield and initial population of node locations with high posterior probability (given data collected at the network startup) and (ii) a sequential Monte Carlo (SMC) algorithm for recursively tracking the target position and velocity and sequentially re-generating new populations of node positions as new observations become available. The resulting algorithm is implemented in a distributed fashion. Assuming that the communication capabilities of the DFCs enable them to share some data, each DFC can run an independent SMC algorithm and produce local estimates of the magnitudes of interest. Optimal data fusion is achieved by a linear combination of the local estimates with adequate weights. We illustrate the application of the algorithm in a network of power-aware sensors.