Low complexity tracking for ad-hoc automotive sensor networks

The main focus of this paper is to investigate how the co-operative nature of an ad-hoc sensor network can be exploited in order to reduce the complexity of accurate node locationing algorithms in sensor networks. We propose a novel approach to target tracking, called node-aided tracking. The node-aided tracking algorithm is assisted in the tracking process by the tracked nodes in the network, allowing for a reduction in filter complexity. This approach is shown to be sensitive to network delays. The effects of such delays are investigated, and simple countermeasures are presented that stabilize the algorithm. Further, we present a novel way of evaluating the performance of tracking filters used in automotive safety applications. This new measure, called the time margin difference, takes not only the mean-squared-error into account but also latency in providing location estimates and other important filter characteristics for a fair comparison between different tracking algorithms designed for automotive safety applications. At the end of the paper we present a brief comparison, based on simulation, between the node-aided tracking filter and a traditional high-order Kalman filter with respect to this new performance measure.