Bias phenomenon and compensation in multiple target tracking algorithms

Bias phenomenon in multiple target tracking has been observed for a long time. This paper is devoted to a study of the bias resulting from the miscorrelation in data association. One result of this paper is a necessary condition for miscorrelation to cause bias. Relying on the necessary condition and a model for data association process, techniques are developed to give general directions for where and how to compensate the bias related to miscorrelation in general tracking algorithms. Case studies on the bias phenomenon in two tracking algorithms, i.e., global nearest neighborhood (GNN) and joint probabilistic data association (JPDA), are launched as a practice of the ideas and results presented in this paper. The outcome of the examples illustrates and strongly supports our results. A discussion of several statistical issues is given in the end of this paper, in which the behavior for the bias in GNN and JPDA is studied.