Distributed Implementation of the Centralized Generalized Labeled Multi-Bernoulli Filter

Distributed scenarios pose a big challenge to tracking and fusion systems. They require the prevention of repeatedly incorporating same information, which originates from ring closures in communication path would affect optimality. Additionally, multi-sensor multi-object Generalized Labeled Multi-Bernoulli filter update is NP-hard principle. The method proposed this paper tackles these problems, as it constitutes divide conquer strategy for distributed, synchronized systems with central fusion. Based on common prediction, local sensor updates are calculated separately, sent back fused centrally order start new cycle. Thus, intractable split into less complex single-sensor novel, low-complexity strategy. enables full parallelization optimal -Generalized update. Our approach bases Bayes Parallel Combination Rule can be seen Information Matrix Fusion synchronous sensors, perfect choice centralized distributed sensors. Finally, we compare Iterator Corrector literature detailed simulations.