Quantum federated learning for autonomous vehicle cybersecurity: An analytical review of architectures and threat landscapes

Abstract

With the rapid rollout of autonomous vehicles (AVs), many AV criminal attacks form one of the biggest cybersecurity attack surfaces in contemporary infrastructure. But there are also the AV subsystems—LiDAR, radar, cameras, GPS, CAN buses, and V2X communications—all of which expose separate exploit vectors that would threaten both passenger safety as well as traffic. Centralized machine learning has been utilised for intrusion detection, but privacy issues, scalability limitations, and susceptibility to quantum-enabled adversaries have motivated interest in Federated Learning (FL) as a distributed approach. On the one hand, vehicles operating on RSA- and ECC-based protocols will become insecure due to quantum effects, motivating the need for Quantum Key Distribution (QKD), Post-Quantum Cryptography (PQC) [7], or any alternative to current systems. This seminal work proposes a summarization of Quantum Federated Learning (QFL) research specifically for AV cybersecurity. The contributions are: (i) a multi-dimensional risk matrix for AV attack priorities; (ii) a cross-study synthesis of gaps in FL-IDS deployment; (iii) comparative analysis across QKD, PQC and hybrid models of quantum security transition; (iv) construction of a QFL reference architecture with seven gaps agenda and (v) structured 2025–2030 road-map. The analysis shows that neither FL nor quantum security of it alone is strictly sufficient, and their togetherness in the hinge point between them — QFL — is the most urgent way forward for securing autonomous vehicles.