Behavioral Utility-based Distributed Detection in the Presence of Byzantines
IEEE Signal Processing Letters, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1109/lsp.2026.3715140
- Dergi Adı: IEEE Signal Processing Letters
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Compendex, INSPEC, Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
- Anahtar Kelimeler: behavioral utility, Byzantine attacks, Distributed detection, likelihood ratio test, prospect theory
- Hacettepe Üniversitesi Adresli: Evet
Özet
In this work, we consider a distributed detection system for a binary hypothesis testing problem with conditionally independent observations at the local decision agents (DAs) that are binary quantized and transmitted to a fusion center (FC). Some of the DAs act as malicious attackers, called Byzantines, with the intention of degrading the overall system performance. Unlike classical distributed detection, the FC is modeled as a human operator whose decisions reflect the nonlinear probability distortions and asymmetric gain-loss valuations as characterized by prospect theory. The FC employs a fixed fusion rule, while Byzantine DAs choose their local decision rules to minimize the FC's behavioral utility and honest DAs choose theirs to maximize it. This leads to a Stackelberg formulation in which the FC commits to its fusion rule, the Byzantine nodes respond adversarially, and the honest nodes optimize accordingly. The optimal Byzantine strategy is shown to exist and characterized to be a randomization between two single-threshold likelihood ratio-based decision rules. Our theoretical results are also corroborated by numerical examples.