Recursive LMMSE Centralized Fusion with Compressed Multi-Radar Measurements
Qingpeng Zhang, Zhansheng Duan, Uwe D. Hanebeck · 2019
For multi-sensor centralized fusion with linear measurements, simply stacking all measurements up and then applying the Kalman filter at the fusion center can give the optimal estimation performance. This optimal performance is independent of how the measurements from different sensors are stacked up. However, for centralized fusion with multiple radar measurements under the recursive LMMSE filtering framework, the performance really matters as to how to stack the measurements from different radars. In [1], we have shown that centralized fusion with stacked recombined multi-radar measurements outperforms the one with stacked original measurements under the recursive LMMSE filtering framework. In this paper, we further develop a new multi-radar centralized fusion approach by compressing all measurements first and then applying the recursive LMMSE filter with single radar measurements at the fusion center. Numerical examples show that the new centralized fusion with compressed measurements has better estimation accuracy and smaller noncredibility than the ones with stacked measurements.