Event-Triggered Consensus LMB Filter for Distributed Multitarget Tracking

Abstract

This correspondence investigates the problem of reducing energy and bandwidth consumption of a sensor network for distributed multitarget tracking by the labeled multi-Bernoulli filter. To this end, an event-triggered method is adopted together with a consensus strategy for each sensor node to transmit only Bernoulli components (BCs) that achieve enough information gain. Two message transmission strategies, i.e., joint and independent transmission, are devised. The possibility to separately trigger BCs allows the proposed method to better tradeoff tracking performance versus communication load with respect to existing distributed multitarget tracking approaches.