A target tracking in wireless sensor networks consists of two main functions: The detection and the tracking of the target along its trajectory by means of sensors deployed in an area of interest. Generally, these sensors are not maintainable after deployments. Dynamic clustering algorithms seem to be an effective mechanism for increasing the network’s lifetime. Indeed, this type of algorithms only activates the nodes that are on the trajectory of the target when the latter is at their reach. All other sensors must be in sleep mode. The effectiveness of a monitoring solution must take into account the quality of monitoring, connectivity, and the power consumption that are directly affected by the distribution and density of the nodes. We propose to construct optimal dynamic clusters on the target trajectory based on a probabilistic model integrating two fundamental parameters: energy consumption and accuracy. This last metric is evaluated, for the first time in the target tracking algorithms, by the notion percolation.

无线传感器网络中的目标跟踪包括两个主要功能：利用部署在感兴趣区域中的传感器对目标沿其轨迹进行检测和跟踪。通常，这些传感器在部署后无法维护。动态集群算法似乎是增加网络寿命的有效机制。实际上，这种类型的算法仅在目标到达时才激活目标轨迹上的节点。所有其他传感器必须处于睡眠模式。监视解决方案的有效性必须考虑到监视的质量，连接性以及受节点分布和密度直接影响的功耗。我们建议基于集成了两个基本参数：能耗和精度的概率模型，在目标轨迹上构造最佳动态聚类。在目标跟踪算法中，最后一个度量是通过概念渗滤首次评估的。