Informative mobility scheduling for mobile data collector in wireless sensor networks

In this paper, we study the issue of mobility scheduling for mobile data collector (MDC) in wireless sensor networks. Most existing work in this area focuses on geometric-based optimization without considering the spatial correlation among different locations. In this paper, we study the mobility scheduling problem from the informative perspective by using Gaussian process to capture the spatial correlation of real world phenomena. Based on the Gaussian process model and collected sensing data from a number of sensor nodes in the network, one can predict the sensing values at the remaining interesting locations and can further estimate the prediction accuracy. This approach can potentially shorten the length of data collection tour with small penalty in data accuracy. We use the mutual information maximization criteria to evaluate the quality of a data collection tour. We accordingly formulate the informative mobility scheduling problem which finds the data collection tour with the maximal mutual information under certain mobility constraint. The problem is shown to be NP-hard and we accordingly propose two efficient heuristic algorithms. We evaluate the performance of our algorithms by comparing them with geometric-based algorithms through extensive simulations and the results show that our algorithms can return much shorter tours while achieving the same level of data quality.