Source-channel coding over Gaussian sensor networks with active sensing

A limited energy budget is a major obstacle to the practical, wide deployment of sensor networks and hence necessitates the judicious optimization of available resources. In this paper, joint optimization of sensing and communication resources to minimize total energy spent within a sensor network is considered. A particular sensor network model with one Gaussian source observed by many sensors, subject to additive independent Gaussian observation noise, is examined. Sensors communicate with the receiver over an additive Gaussian multiple access channel. The aim of the receiver is to reconstruct the underlying source with minimum mean squared error. The fundamental tradeoff between communication and sensing over this sensor network model is characterized. Under symmetric conditions, for a single sensor, power is shared equally between communication and sensing. As the number of sensors increases, the sensing error dominates the overall error expression, hence sensing takes almost all power. The optimal power scheduling among sensors in the asymmetric case is determined, and it is shown that the power allocation schedule admits a simple decentralized implementation. Numerical results show that joint optimization of communication and sensing power yields significant power savings compared to the conventional approach of optimization of only communication power allocation.