区间聚类结合区域划分法实现WSN随机游走模型跟踪
A Solution to WSN Random Walk Model Tracking Based on Interval Clustering and Area Division
-
摘要: 传统的移动目标跟踪方案大都基于理想运动模型,利用滤波方法(如Kalman滤波)进行连续的位置估计和预测。但现实世界中,目标移动规律往往带有强机动性,无任何规律可言,称为随机游走模型。基于此,本文提出了一种低复杂度解决方案,引入离线标准参考样本空间替代在线位置估测过程,为移动目标提供了一种启发式跟踪方案。仿真结果表明,该算法在保证精度的同时降低了计算复杂度,对系统能耗设计有一定指导意义,同时具有更高的灵活性和实用性。Abstract: Most of traditional tracking algorithms are based on ideal models and achieve continuous position estimates and predictions by using filtering methods. However, in real-world, moving targets often have no patterns, termed as the random walk model. Hence, this paper proposes a lower-complexity solving scheme for this kind of disorder movement model. By changing positon tracking problem into a data sequence matching problem, the proposed algorithm provides a heuristic tracking scheme for random moving targets. Numerical simulation experiments show that this scheme can effectively reduce the complexity while ensuring tracking accuracy. Meanwhile, it has greater flexibility and universality.