无线传感器网络目标跟踪动态簇成员调度策略

摘要

By analyzing the inherent relationship among the target tracking monitoring precision, node energy consumption and cluster member wake/sleep in WSNs, partially observable Markov decision process ( POMDP) theory is imported, a dynamic cluster member scheduling model based on node energy consumption weighted reward rate and target tracking accuracy is proposed to address the problem of limited energy, monitoring data redundancy from densely deployed node,sensor node position estimation error and the target estimation error. An online value-iteration policy solving algorithm based on belief point is used to address the problem of high complexity of dynamic cluster-members scheduling. The algorithm realizes dynamic generation and online adjusting of sensor cluster-member nodes cooperation strategy. Simulation results show that this algorithm can improve target tracking accuracy,reduce node energy consumption and prolong network lifetime.%通过分析目标跟踪无线传感器网络监测精度、节点能量消耗与簇成员唤醒/休眠之间的内在联系,针对网络节点能量有限、密集部署节点监测数据存在冗余、传感器节点的自身位置估计误差和目标监测估计误差等问题,引入部分可观察Markov决策过程(POMDP)理论,提出一种基于目标跟踪准确度和节点能量消耗加权回报率的动态簇成员调度模型；针对动态簇成员调度算法复杂度偏高的问题,采用基于信念点的值迭代在线策略求解算法,实现传感器簇成员节点协作策略的动态生成和在线调整.仿真结果表明:该算法能够提高目标跟踪准确性,降低节点能量消耗,延长网络生存时间.