On Optimal Clustering in Mobile Wireless Sensor Networks Under Uncertainty

Wireless sensor networks are comprised of geographically distributed sensors that collect and convey time-critical information without human intervention, making them particularly useful for military applications; however, network performance may be limited by the typically small energy supplies of the sensors. One strategy for dealing with limited energy storage in such networks is to allow sensor nodes to aggregate sensed data at particular nodes known as cluster heads. We examine the problem of optimally locating and relocating cluster heads over a finite time horizon. The objective is to simultaneously maximize the demand coverage and minimize the costs of relocating cluster heads. Our main contribution is to consider explicitly the random evolution of mobile sensor nodes, which requires periodic updating of their precise locations. Computational experiments illustrate the usefulness of such dynamic information updates to help mitigate the uncertainty in sensor node locations and ensure reliable demand coverage by cluster heads.