Energy-driven detection scheme with guaranteed accuracy

This is our first step towards a holistic investigation of the minimum energy for wireless sensor network (WSN) to perform a specific function. We consider wireless sensor networks that perform an event detection function. Each sensor node will repetitively collect a 1-bit information regarding whether the event occurs or not in its neighborhood. A fusion center will make the decision on whether the event occurs based on the information provided by individual sensor nodes. Traditionally, a centralized scheme requires each sensor node to forward all its observations to the fusion center, which results in large energy in communication. A distributed scheme, on the other hand, allows each sensor node to make its own decision and then send out only its 1-bit decision. This reduces communication energy at the cost of increased processing energy and reduced detection accuracy. We propose a hybrid energy-driven scheme where each sensor node sends out its 1-bit decision if that decision exceeds a pre-determined detection accuracy threshold, and sends out all its observations otherwise. This scheme provides WSN designers the flexibility to balance detection accuracy, sensor density, and energy consumption. We develop the optimal decision rules for this scheme. We also propose methods to calculate the detection accuracy threshold for individual sensor node to guarantee the overall detection accuracy at the fusion center. The simulation results show that the hybrid scheme consumes significantly less energy than both centralized and distributed schemes to achieve the same detection accuracy