A general packet dropout compensation framework for optimal prior filter of networked multi-sensor systems

This paper proposes a new packet dropout compensation framework for networked multi-sensor systems. It is more general, which includes the existing popular mechanisms such as the zero-input and hold-input mechanisms as the special cases. Based on the proposed compensation framework, the centralized fusion linear optimal full-order prior filters with compensators of different weighting factors are presented. However, it is well known that centralized fusion prior filters (CFPFs) have poor reliability. To improve the reliability, the distributed fusion prior filters (DFPFs) are also given based on the well-known matrix-weighted fusion estimation algorithm in the linear unbiased minimum variance (LUMV) sense. Moreover, the stability and steady-state property of the proposed CFPFs and DFPFs are analyzed. A numerical example is used to make the performance comparisons of the proposed and the existing compensation mechanisms. The simulation results verify the effectiveness of the proposed compensation framework and the corresponding fusion filters.