Comparison of decentralized tracking algorithms
Ng Gee Wah, Rong Yang · 2003
There are various algorithms in decen- tralized tracking. These algorithms can be categorized into three classes based on the inputs, namely track fusion, measurement fusion and information fusion. The track fusion combines individual tracks formed by different sensors. The measurement fusion generates tracks from the combined measurements detected by dif- ferent sensors. The information fusion is derived from information filter performing estimation and tracking on information space. In the track fusion, the al- gorithms include the simple convex combination, the Bar-Shalom/Campo state vector combination, the best linear unbiased estimation(B1 UE) and the covariance intersection (CI). This paper presents the comparison results of the main fusion algorithms based on the sim- ulation data. The paper also discusses the issue on the communication cost of the fusion algorithms.