摘要： 针对目标追踪无线传感器网络节点能量有限、感知信息存在不确定性等问题，提出一种基于部分可观察马尔可夫决策过程的在线节点调度算法。通过状态转移函数和观察函数描述移动目标的不确定性，根据奖赏函数平衡追踪性能和节点能量消耗，并构造有限深度的可达信念与或树降低运算复杂度，实现调度策略在线求解。实验结果表明，该算法能平衡目标追踪质量与节点能量消耗，且满足实时性 要求。
关键词:
目标追踪,
无线传感器网络,
部分可观察马尔可夫决策过程,
节点调度,
可达信念与或树,
在线算法
Abstract: A Partially Observable Markov Decision Process(POMDP)-based online node scheduling algorithm is proposed to address the problem of limited energy supply and the uncertainty of perception in target tracking Wireless Sensor Network(WSN). The algorithm represents the uncertainty of moving target using state transition and observation function, and uses reward function to balance the tracking performance and node energy consumption. In order to solve the scheduling policy online, it constructs reachable belief AND/OR tree with limited depth to reduce computational complexity. Experimental results show that this algorithm can achieve a good tradeoff between the target tracking quality and the node energy consumption while sufficing real-time demand.
Key words:
target tracking,
Wireless Sensor Network(WSN),
Partially Observable Markov Decision Process(POMDP),
node schedule,
reachable belief AND-OR tree,
online algorithm
中图分类号:
冯延蓬, 仵博, 郑红燕, 孟宪军. WSN中一种目标追踪在线节点调度算法[J]. 计算机工程, 2012, 38(11): 96-99,103.
FENG Yan-Peng, WU Bo, ZHENG Gong-Yan, MENG Xian-Jun. Target Tracking Online Node Schedule Algorithm in Wireless Sensor Network[J]. Computer Engineering, 2012, 38(11): 96-99,103.