Distributed Filtering with Value of Information Censoring

TL;DR

This paper introduces a distributed Bayesian filtering algorithm that reduces communication costs by censoring low-VoI estimates, maintaining accuracy and consistency in dynamic networks.

Contribution

It proposes a VoI-based censoring scheme within a distributed Information filter, balancing communication efficiency and estimation accuracy.

Findings

Significant reduction in communication costs demonstrated in simulations.

Algorithm maintains consistent estimates under linear-Gaussian models.

Validated performance through real-world wireless network experiments.

Abstract

This work presents a distributed estimation algorithm that efficiently uses the available communication resources. The approach is based on Bayesian filtering that is distributed across a network by using the logarithmic opinion pool operator. Communication efficiency is achieved by having only agents with high Value of Information (VoI) share their estimates, and the algorithm provides a tunable trade-off between communication resources and estimation error. Under linear-Gaussian models the algorithm takes the form of a censored distributed Information filter, which guarantees the consistency of agent estimates. Importantly, consistent estimates are shown to play a crucial role in enabling the large reductions in communication usage provided by the VoI censoring approach. We verify the performance of the proposed method via complex simulations in a dynamic network topology and by experimental validation over a real ad-hoc wireless communication network. The results show the validity of using the proposed method to drastically reduce the communication costs of distributed estimation tasks.