Particle filter implementation of the multi-Bernoulli filter for superpositional sensors

The multi-Bernoulli filter is a promising method for computationally efficient and accurate multi-target tracking. Computationally tractable approximations of the multi-Bernoulli filter equations for superpositional sensors were recently derived. In this paper we present a particle filter implementation of these approximate update filter equations. We describe how the filter could be employed to address the radio-frequency tomographic tracking task and conduct a simulation study to compare performance with the probability hypothesis density (PHD) and cardinalized probability hypothesis density (CPHD) filters.