Distributed Sparse Signal Detection With Energy-Efficient Censoring-Quantization in Wireless Sensor Networks

In wireless sensor networks (WSNs), the energy consumption of sensors and the system performance are inevitably contradictory. For the distributed detection problem in WSN, the censoring technique has been proposed as an energy-efficient transmission strategy. However, most current detection schemes with censoring do not consider the quantization operations in actual communication systems to further save the transmission energy in actual communication systems. In this paper, we propose two energy-efficient detection schemes including a common censoring-quantization strategy and two different detectors. This censoring-quantization strategy aims to reduce the energy consumption of transmitting observations. It is modeled as a three-state equivalent transmission strategy at sensors, with the states being keeping silent, transmitting “1”, and transmitting “−1”. To fully explore the advantage of this strategy, we design two detectors, namely the Censoring-Quantization Maximum-Likelihood (CQ-ML) detector and the Censoring-Quantization Locally Most Powerful Test (CQ-LMPT) detector, respectively for two scenarios, which correspond to whether the fusion center (FC) knows the parameters of the Phenomenon of Interest (POI). We further analyze the theoretical performance of our schemes to provide the optimal censoring-quantization thresholds. Finally, experimental results demonstrate the effectiveness of our schemes and highlight the significant energy consumption reduction with negligible performance loss.