Information Exchange Track-Before-Detect Multi-Bernoulli Filter for Superpositional Sensors
Elinor S. Davies, Ángel F. García‐Fernández
Abstract
In this paper we derive the Information Exchange track-before-detect Multi-Bernoulli (IEMB) filter for multi-target filtering with superpositional sensors. The IEMB filter propagates a multi-Bernoulli density through the filtering recursion and each Bernoulli is propagated with its own prediction and update step. At each update step, each Bernoulli filter exchanges the predicted mean and covariance matrix of its measurement contribution with the other Bernoulli filters. The exchanged information is then used by the filters to perform the update step. Additionally, we propose the Iterated Posterior Linearisation Filter (IPLF) implementation of the IEMB filter (IEMB-IPLF). We compare the IEMB-IPLF filter to a number of other non-linear filtering methods showing the benefits of the proposed filter.