Tracking with Unreliable Node Sequences

Abstract

Tracking mobile targets using sensor networks is a challenging task because of the impacts of in-the-flled factors such as environment noise, sensing irregularity and etc. This paper proposes a robust tracking framework using node sequences, an ordered list extracted from unreliable sensor readings. Instead of estimating each position point separately in a movement trace, we convert the original tracking problem to the problem of finding the shortest path in a graph, which is equivalent to optimal matching of a series of node sequences. In addition to the basic design, multidimensional smoothing is developed to enhance tracking accuracy. Practical system deployment related issues are discussed in the paper, and the design is evaluated with both simulation and a system implementation using Pioneer III Robot and MICAz sensor nodes. In fact, tracking with node sequences provides a useful layer of abstraction, making the design framework generic and compatible with different physical sensing modalities.