Lightweight Noise Diagnosis for Photonic Quantum Detectable Byzantine Agreement
Kevin Bogner, Kuan‐Cheng Chen, Aysajan Abidin, Kin Kwong Leung · 2026
Detectable Byzantine Agreement (DBA) protocols abort rather than risk inconsistent decisions when verification fails. Quantum variants (QDBA) implement this detectability via multipartite entanglement checks, but in photonic networks nonuniform noise across participants makes these checks brittle: a handful of noisy nodes can trigger needless aborts or conceal adversarial behavior. We introduce two drop-in diagnosis methods that run during the standard entanglement-verification step of QDBA protocols: (i) a batch health check that detects both state leakage and insufficient fidelity, and (ii) a per-node diagnosis that estimates each participant's error rates and labels loss-biased outliers. The methods require no extra quantum rounds and add only modest classical overhead. In simulations, the batch health check substantially reduces false accepts while preserving cleanbatch acceptance, and the per-node diagnosis reliably flags noisy Lieutenants and the Commander at practical verification sizes. The techniques are compatible with existing QDBA protocols.