Performance-Energy Tradeoffs for Decentralized Estimation in a Multihop Sensor Network

The performance-energy tradeoffs for decentralized estimation in a multihop sensor network are studied. The sensor network consists of multiple heterogeneous branches. The leaf node on each branch observes a parameter corrupted by noise. Each leaf node quantizes its observation and sends the quantized data to the fusion center hop by hop. The parameter is estimated at the fusion center by using the best linear unbiased estimator. The performance is evaluated by adopting the MSE criterion. The total energy consumption in the sensor network is calculated. Network lifetime is defined and analyzed under per-sensor energy constraint. The tradeoff between performance and energy consumption as well as tradeoff between performance and network lifetime are investigated simultaneously. Optimization problems are formulated regarding both tradeoffs, and the conditions for optimization are identified for specific and generic sensor network deployments. The results demonstrate that both tradeoffs exist under given conditions.