OmniGuard V2X: A Hybrid-Security Prototype Framework with Assumption-Aware Validation for Smart Vehicle Systems

Abstract

Abstract Vehicle-to-everything (V2X) systems combine safety-relevant telemetry, wireless communication, distributed identity, collaborative learning, and long-lived cryptographic trust, creating security dependencies that are difficult to evaluate when each mechanism is studied in isolation. This paper presents OmniGuard V2X, a six-layer research prototype that integrates post-quantum key establishment and signatures, privacy commitments, decentralized identity, a permissioned tamper-evident ledger, federated-learning experimentation, anomaly hooks, and authenticated runtime services in a unified C++/Go/Python stack. The contribution is an assumption-aware validation model rather than a new cryptographic primitive. The supported native path uses real ML-KEM-512 and ML-DSA-44 operations, while classical Pedersen and Schnorr-style assumptions are explicitly excluded from end-to-end post-quantum claims. The prototype further implements encrypted persistent PQC identity material, stable key identifiers, guarded rotation, signed key-transition evidence, bounded historical signature verification, fail-closed provider selection, replay-aware local control, and authenticated rollback-state checks. A retained warm-start pipeline measurement is 5.34 ms in one prototype environment and is reported only as a descriptive local result. Existing anomaly and federated-learning experiments are classified as sanity checks because they do not yet support statistically defensible robustness metrics. The study shows how machine-readable security boundaries and evidence-aware claim control can improve the reproducibility and interpretability of multi-mechanism V2X security prototypes.