Adaptive quantization for multihop progressive estimation in wireless sensor networks

We consider the problem of parameter estimation in a wireless sensor network, where because of the bandwidth and power constraints, each sensor transmits quantized information to its parent on a multihop path. Our approach jointly optimizes: i) sensor selection, ii) routing structure, and iii) number of bits per sample for each sensor. First, we express our problem as an optimization problem, and then we design an algorithm that is based on an adaptive quantization and an estimate-and-forward scheme that allows performing sequentially this joint optimization in an efficient way. We show that our algorithm provides a routing structure that trades-off the aforementioned three metrics better than the traditional shortest path tree routing structure, which is commonly used in practice. Numerical results depicting the performance and advantages of our approach over previous state-of-the-art approaches are presented.