Title: Federated learning and dynamic game-based collaborative optimisation for resource allocation in IoT data acquisition

Authors: Yunqi Wang; Liangliang Ding

Addresses: School of Electronics and Information Engineering, Suzhou Polytechnic University, Suzhou, 215104, China ' Service Facilities Division, Taicang Highway Development Centre, Taicang, 215400, China

Abstract: Internet of things networks are expanding rapidly, making efficient resource allocation for data acquisition a critical challenge. The resources considered include communication bandwidth, energy, and computational capabilities. Traditional centralised optimisation methods face significant difficulties due to limitations in these resources, as well as privacy concerns. This paper proposes a collaborative optimisation framework combining federated learning and dynamic game theory to achieve decentralised and adaptive resource allocation in IoT data acquisition systems. The approach enhances privacy protection while reducing communication overhead. Existing federated learning methods have shown reductions in communication costs - specifically, the number of communication rounds and data volume - by up to 94.89%. Dynamic game approaches in IoT have demonstrated improvements, including 42% higher packet delivery ratios and up to 32% lower latency, in environments with moderate node density and interference levels. The proposed framework helps balance the growing energy demands of IoT networks while ensuring data security and transmission efficiency.

Keywords: federated learning; FL; dynamic game theory; resource allocation; internet of things; IoT; data acquisition; collaborative optimisation; privacy preservation.

DOI: 10.1504/IJSNET.2026.153107

International Journal of Sensor Networks, 2026 Vol.50 No.4, pp.273 - 284

Received: 25 Sep 2025
Accepted: 30 Sep 2025
Published online: 22 Apr 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article