Stochastic Network Collection Point (NCP) Selection in Mobile Sensor Networks with Cooperative Communications

In wireless sensor networks (WSNs) with mobile sinks, some stationary sensor nodes, known as network collection points (NCPs), can communicate with mobile sinks to collect sensed data. Most existing works on mobile WSNs assume that mobile sinks directly communicate with NCPs via point-to-point links Consequently, recent advances in cooperative communications, which are more energy-efficient than traditional point-to-point communications, are largely ignored in mobile WSNs. In this paper, we present a novel stochastic NCP selection scheme using cooperative communications. In addition, most existing works assume perfect channel state information (CSI) knowledge, which may not be realistic in practice. We only assume inaccurate CSI in this paper. We formulate the problem as a stochastic multi-armed bandit system. This formulation facilitates the distributed architecture due to its "indexibility" characteristic, which dramatically simplifies the computation and implementation of the proposed scheme. Simulation results are presented to show the effectiveness of the proposed scheme.