An incentive-aware federated bargaining approach for client selection in decentralized federated learning for IoT...

Federated Learning (FL) has emerged as a promising solution for privacy-preserving model training across distributed IoT devices. Despite its advantages, FL faces challenges such as inefficient client selection, data heterogeneity, security vulnerabilities, and exposure to Man-in-the-Middle (MITM) attacks. To address these issues, the Incentive-Aware Federated Bargaining (IAFB) framework is proposed, integrating Nash Bargaining for optimal client selection, Shapley-value-based incentives for fair reward distribution, and decentralized peer-to-peer (P2P) aggregation to eliminate single points of failure. To enhance security, IAFB employs AES-GCM encryption, ensuring data confidentiality, authenticity, and integrity during transmission, effectively mitigating MITM attacks. Experimental results demonstrate that IAFB improves participation fairness by 28%, boosts model accuracy by 6.5%, and reduces convergence time by 35% compared to FedAvg. Additionally, IAFB reduces communication overhead by 39.5% and enhances resilience against adversarial threats, making it highly suitable for secure and scalable FL deployment in resource-constrained IoT environments.