MIRACLE: Mobility Prediction Inside a Coverage Hole Using Stochastic Learning Weak Estimator

In target tracking applications of wireless sensor networks (WSNs), one of the important but overlooked issues is the estimation of mobility behavior of a target inside a coverage hole. The existing approaches are restricted to networks with effective coverage by wireless sensors. Additionally, those works implicitly considered that a target does not change its mobility pattern inside the entire tracking region. In this paper, we address the above lacunae by designing a stochastic learning weak estimation-based scheme, namely mobility prediction inside a coverage hole (MIRACLE). The objectives of MIRACLE are two fold. First, one should be able to correctly predict the mobility pattern of a target inside a coverage hole with low computational overhead. Second, if a target changes its mobility pattern inside the coverage hole, the proposed estimator should give some estimation about all possible transitions among the mobility models. We use the trajectory extrapolation and fusion techniques for exploring all possible transitions among the mobility models. We validate the results with simulated traces of mobile targets generated using network simulator NS-2. Simulation results show that MIRACLE estimates the mobility patterns inside coverage hole with an accuracy of more than 60% in WSNs.