Distributed Particle Filter Based on Particle Exchanges

Distributed Particle Filter Based on Particle Exchanges

Tang, Rui (Student TU Delft)
Riemens, E.H.J. (TU Delft Signal Processing Systems)
Rajan, R.T. (TU Delft Signal Processing Systems)

2023

Multi-agent networks are known for their scalability, robustness, flexibility, and are typically tasked with a variety of tasks such as target tracking, surveillance, traffic control, and environmental monitoring. Distributed Particle Filters (DPF) are often employed when the for non-linear parameter estimation with non-Gaussian noise. In this paper, we propose a novel distributed particle filter whose transmitted quantities are particles. The fusion process of particles is implemented in a distributed and iterative fashion. To reduce the communication overhead, we adopt the Gaussian process-enhanced resampling algorithm, which reduces the size of local particle set, while still ensures acceptable filtering performance. To determine the local particle set after the communication, we propose two solutions. Our first algorithm (GP-DPF) adopts a “scoring mechanism”, allowing local agents score the received particles and using the scores as the selection criterion. Our second proposed solution (FA-DPF) is a meta-heuristic approach, which uses the well known firefly algorithm as a selection method for particle-based distributed particle filtering. Our simulations demonstrate the superiority of our proposed algorithms under the condition of limited communication and computational resources against other state-of-the-art distributed particle filters.

http://resolver.tudelft.nl/uuid:2ccf6a1b-2b53-4e99-b863-999c2173ff5c

https://doi.org/10.1109/AERO55745.2023.10115781

IEEE

2023-11-15

978-1-6654-9033-7

Proceedings of the 2023 IEEE Aerospace Conference

2023 IEEE Aerospace Conference, 2023-03-04 → 2023-03-11, Big Sky, United States

IEEE Aerospace Conference Proceedings, 1095-323X, 2023-March

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Institutional Repository

conference paper

© 2023 Rui Tang, E.H.J. Riemens, R.T. Rajan