Neighbor-Corroborated Kalman Filtering for Robust Distributed Multi-Object Tracking

Read full text Summarise with VIRAT
Authors: Vahid Ghorbani, Amirali Khodadadian Gostar, Reza Hoseinnezhad, Zahir Tari, Amir Ghorbani, Giorgio Battistelli
Cited by: 0
References: 20

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Robust multi-object tracking (MOT) in distributed sensing networks is increasingly challenged by adversarial manipulation of measurements. In particular, stealthy ghost attacks generate observations that mimic genuine targets, making them difficult to distinguish using conventional trackers. While the ALARM (Average Likelihood for Attack-Resilient Multi-object) framework has recently been introduced within the Random Finite Set (RFS) paradigm, many practical systems rely on computationally efficient Kalman-based pipelines. This paper bridges this gap by proposing KF-ALARM, a distributed filtering framework that embeds the ALARM principle into the Kalman update. The method leverages cross-node track corroboration to modulate the measurement likelihood, attenuating the influence of adversarial measurements. The resulting approach achieves attack-resilient distributed MOT while preserving the efficiency and scalability of Kalman-based tracking architectures.

Fields of Study & Subjects

Target Tracking and Data Fusion in Sensor Networks Video Surveillance and Tracking Methods Adversarial Robustness in Machine Learning

Publication details

Publication date
30 March 2026
Language
EN
Document type
Preprint