基于观测节点不等概随机投影的压缩感知

Compressed Sensing Based on Random Projection of Observation Nodes with Unequal Probabilities

基于空间相关性的压缩感知是无线传感器网络的一个重要研究内容,采用随机选取节点的方法构建观测矩阵可以有效降低观测节点的能量损耗。提出一种基于不等概随机投影的节点选择策略,每个节点根据采集数据和门限的比较结果选择高概率或低概率,并根据能量均衡策略对其进行调节后以该概率发送数据至融合中心。仿真结果表明,与等概率随机投影相比,文章提出的策略在不增加额外能量损耗的前提下,降低了整个区域重构误差。

Compressed sensing based on temporal and spatial correlation is one important research field in wireless sensor network（ WSN）. Making the measurement matrix through random nodes selection can effectively reduce the energy consumption of observation nodes. A strategy to select nodes based on random projection with unequal probabilities is proposed. Each node chooses a higher probability or a lower one according to the comparison between its measurement and a threshold.It revises the probability according to an energy balance strategy and sends its observation to the fusion center with the revised probability. Simulation results demonstrate that compared with random projection with equal probability,the proposed strategy can lower recovery errors both in the whole fields and the interesting fields without extra energy consumption.