Federated Reinforcement Learning for Intelligent Traffic Signal Control: A Privacy-Preserving Approach with Edge-Assisted Aggregation

Ch. 159. P. 1–5.

Ali J. Dayoub, Ehab S. Suleiman

Abstract— Urban traffic congestion costs the global economy over $1 trillion annually, necessitating intelligent traffic signal control (ITSC) solutions. Traditional centralized approaches face critical limitations: privacy violations from vehicle trajectory data sharing, prohibitive communication overhead, and scalability challenges in heterogeneous urban environments. This paper presents a federated reinforcement learning (FRL) framework for privacy-preserving traffic signal optimization. The proposed approach combines three innovations: edgeassisted aggregation weighting client contributions by local traffic density, FedProx regularization handling non-IID traffic distributions, and differential privacy. Evaluated on a 4×4 urban grid using SUMO simulation with four federated clients, our framework achieves 86.5% improvement in waiting time over 100 rounds, outperforms centralized Deep Q-Networks (DQN) by 56.6% at round 50 despite avoiding raw data access, and reduces communication overhead by 8,333. Results establish federated learning as a viable paradigm for scalable, privacy-compliant intelligent transportation systems with
superior performance to centralized approaches.

Keywords: Intelligent transportation systems, edge computing, Federated learning, traffic signal control

In book

Proceedings of the 2026 8th International Youth Conference on Radio Electronics, Electrical and Power Engineering (REEPE)

Dayoub A., Suleiman E. IEEE, 2026.