Adaptive quantization for distributed estimation in cluster-based wireless sensor networks

In this paper, the problem of parameter estimation in cluster-based wireless sensor networks is studied. Particularly, we focus on how to choose a suitable threshold in the one-bit adaptive quantization scheme. An adaptive quantization scheme for parameter estimation in cluster-free sensor networks is extended to this scenario. Intra-cluster and inter-cluster maximum likelihood estimators (MLEs) as well as the corresponding Crame´r–Rao lower bounds (CRLBs) are derived. Due to the energy constraint of sensors, the performance–energy tradeoff of parameter estimation is also investigated. Simulation results show that parameter estimation in cluster-based sensor networks with adaptive quantization can be more energy-efficient than that in cluster-free sensor networks, while achieving close performance as the number of sensors increases.