The exact algorithm for multi-sensor asynchronous track-to-track fusion

Track-to-track fusion is an important topic for distributed tracking. Compared with the centralized measurement fusion, the track-to-track fusion requires less communication resources and is suitable for practical implementation. Although having been widely investigated in the literature, the majority of track-to-track fusion algorithms assume synchronous communication. However, in practice, local sensors might communicate to the fusion center with arbitrary communication intervals, which raises the problem of multi-sensor asynchronous track-to-track fusion. In this paper we develop an exact fusion algorithm to solve this problem, under the condition that at each time step only parts of the sensors send their local estimation tracks to the fusion center. Formulas are derived to obtain the exact cross-covariances between the local tracks by taking into consideration the impact of the potential feedback from the fusion center. Based on the derived formulas, the scalable fusion algorithm is developed and validated with extensive Monte Carlo simulations.