On Arithmetic Average Fusion and Its Application for Distributed
Multi-Bernoulli Multitarget Tracking
Abstract
Recently, the simple arithmetic averages (AA) fusion has demonstrated
promising, even surprising, performance for multitarget information
fusion. In this paper, we first analyze the conservativeness and Frechet
mean properties of it, presenting new empirical analysis based on a
comprehensive literature review. Then, we propose a target-wise fusion
principle for tailoring the AA fusion to accommodate the multi-Bernoulli
(MB) process, in which only significant Bernoulli components, each
represented by an individual Gaussian mixture, are disseminated and
fused in a Bernoulli-to-Bernoulli (B2B) manner. For internode
communication, both the consensus and flooding schemes are investigated,
respectively. At the core of the proposed fusion algorithms, Bernoulli
components obtained at different sensors are associated via either
clustering or pairwise assignment so that the MB fusion problem is
decomposed to parallel B2B fusion subproblems, each resolved via exact
Bernoulli-AA fusion. Two communicatively and computationally efficient
cardinality consensus approaches are also presented which merely
disseminate and fuse target existence probabilities among local MB
filters. The accuracy and computing and communication cost of these four
approaches are tested in two large scale scenarios with different sensor
networks and target trajectories.