A Multisource Multi-Bernoulli Filter for Multistatic Radar

Abstract

Compared with conventional monostatic or bistatic radar, multistatic radar has wider coverage, better performance of localization and higher tracking accuracy. However, the multistatic radar architecture poses challenges to the implementation for multitarget tracking in coping with highly uncertainty of data association for the fusion of multisource information. In this paper, the theoretically rigorous formulas for the multisource multi-Bernoulli (MeMBer) filter are derived by using the Finite set statistics (FISST) calculus built on the standard MeMBer filter. The multisource MeMBer filter propagates a set of MeMBer parameters approximately characterizing the multisource corrected posterior multitarget random finite set (RFS). Since the equations for the proposed filter multisource corrector are computationally intractable, we go further to develop an analytic Sequential Monte Carlo (SMC) implementation of multisource MeMBer recursion. The theoretical analysis and simulations show that the proposed filter performs well and accommodates nonlinear multistatic radar tracking scenario with a single transmitter and two receivers under the approximate conditions.

Keywords

• Multistatic radar
• finite set statistics
• multi-Bernoulli filter
• multitarget tracking