Dynamic architectural reconfigurations of sensor networks

We consider sensor networks with a specific signal processing objective. The networks are organized in architectures comprised of sensor clusters whose cluster heads are connected via a backbone network. The data collected by the sensors are finally fused at a fusion center to satisfy the designated signal processing objective. Data operations and their time limitations are dictated by the signal processing objective, in conjunction with the power and life-span constraints of the sensors. The limited life-span of the sensors induce time-varying cluster traffic rates, and, thus dynamics in the operation of any rate allocation schemes, while they may also necessitate dynamically evolving network architectures. In this paper, we focus on network dynamic architectural reconfigurations, as dictated by time-varying aggregate traffic rates in each cluster. The traffic rate changes are detected dynamically by a traffic monitoring algorithm with powerful characteristics.