Noise-resilient Quantum Federated Learning Enables Low-Latency ADAS Training on NISQ

Advanced Driver Assistance Systems increasingly rely on collaborative learning, but conventional methods struggle with the noisy, delayed and insecure conditions of real-world vehicular networks.

Chethana Prasad Kabgere and Sudarshan T S B, along with their colleagues, address this challenge by introducing a new framework, Noise-Resilient Quantum Federated Learning, which combines the strengths of both quantum and classical computing. This hybrid approach encodes information using quantum states and employs specialised circuits that maintain accuracy even with the imperfections of current quantum hardware.

The team demonstrates that their method not only improves the speed and efficiency of model training, but also enhances security and stability, paving the way for more reliable and intelligent driver assistance systems at the edge of the network. This novel framework combines quantum and classical computing to enable collaborative model training across vehicles while preserving data privacy and mitigating the challenges of real-world network conditions. Experiments reveal that NR-QFL achieves a peak accuracy of 86. 1% and an F1-score of 0. Communication costs were minimized, with NR-QFL incurring only an 8% increase relative to QFL, despite the significant gains in accuracy and stability. Detailed analysis using Class Activation Maps (CAMs) demonstrates that NR-QFL preserves sharper and more coherent feature attention compared to both FedAvg and QFL, indicating improved perceptual integrity, a critical requirement for safety-critical ADAS applications. Simulations indicate that a modular quantum co-processor, utilizing superconducting transmon qubits, can achieve secure, low-latency, and noise-resilient federated learning directly at the vehicular edge, with aggregation latency below 10 milliseconds. High-throughput interconnects, such as PCIe Gen4, provide a 2. 5x improvement in bandwidth, further reducing latency and enabling real-time performance. This approach addresses key challenges in federated learning, such as noise, latency, and security concerns inherent in real-time vehicular networks. By encoding model parameters as quantum states and utilizing variational quantum circuits, NR-QFL achieves robust and reliable convergence even under noisy conditions commonly found in vehicular environments. Empirical validation confirms that NR-QFL delivers measurable accuracy and robustness gains, establishing a scalable and technically feasible foundation for next-generation intelligent systems at the vehicular edge. Future research directions include exploring quantum, classical co-design at the pulse level, investigating thermal-aware hardware integration, and extending the framework to support more complex quantum models, aiming to further enhance energy efficiency and pave the way for fully quantum-enhanced ADAS capabilities. 👉 More information 🗞 Noise-Resilient Quantum Aggregation on NISQ for Federated ADAS Learning 🧠 ArXiv: https://arxiv.org/abs/2512.13196 Tags: Rohail T. As a quantum scientist exploring the frontiers of physics and technology. My work focuses on uncovering how quantum mechanics, computing, and emerging technologies are transforming our understanding of reality. I share research-driven insights that make complex ideas in quantum science clear, engaging, and relevant to the modern world. Latest Posts by Rohail T.: Numerical Solutions Advance Understanding of Tides in Massive Binaries with 0.5 Day Periods December 17, 2025 High-purity Frequency-Degenerate Photon Pairs Advance Scalable Quantum Information Processing December 17, 2025 Coulomb Crystallization Advances Control of Xenon Highly Charged Ions in Traps December 17, 2025