Distributed multiple target tracking and data association in ad hoc sensor networks
FUSION 2003: PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE OF INFORMATION FUSION, VOLS 1 AND 2(2003)
摘要
We introduce an efficient distributed algorithm for tracking multiple targets in an ad hoc sensor network. The key idea is to explicitly model an intermediate representation, which we refer to as individual track contributions, so that the effect of measurements observed by a node of the sensor network can be incorporated locally. Furthermore, since targets affect the observations of a local neighborhood of sensor nodes, communications are limited to local regions around currently known targets. A particularly important problem in multiple target tracking is the problem of data association. Since a sensor network is capable of measuring heterogeneous data, we divide the possible resolutions of ambiguity in data association into three categories. Discriminatory information is: (1) immediately available to resolve the ambiguity in data association, (2) delayed to resolve the ambiguity, and (3) unavailable. In this paper, we demonstrate the incorporation of target identity information when discriminatory information is available and a factored representation of tracks when discriminatory information is unavailable. We build efficiency into the tracking algorithm by triggering the extraction of discriminatory information and triggering the use of such information during tracking only when it is deemed necessary.
更多查看译文
关键词
multiple target tracking,classification,data association,sensor network,distributed algorithm
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络