Work

A Novel PODMAI Framework Enhanced by User Demand Prediction for Resource Allocation in Spectrum Sharing UAV Networks

Year: 2025

Type: article

Abstract: Spectrum sharing unmanned aerial vehicle (UAV) network is a promising technology for future communication systems to mitigate the spectrum scarcity problem. However, the future sixth-generation large-scale wireless communication networks are expected not only to provide a high data rate for massive numbers of users but also to meet their stringent service requirements. Particularly in dynamic spectrum sharing UAV networks, the coupling of multi-dimensional resources and diverse user demands make the efficient and real-time resource allocation exceptionally challenging. A partially observable deep multi-agent active inference (PODMAI) framework is proposed to tackle these issues. The variational free energy is minimized to update the policy exploiting the belief based learning method. A decentralized training and execution multi-agent strategy is designed to navigate the challenges posed by partially observable information. To further satisfy the dynamic user demand and supplement partial observations, a joint spatial-temporal-attention prediction network is designed to construct the demand prediction enhanced PODMAI framework for resource allocation. Exploiting the established framework, an intelligent spectrum allocation and trajectory optimization scheme is elaborated for a spectrum sharing UAV network with multi-modal dynamic transmission rate demands. Simulation results demonstrate that our proposed scheme outperforms benchmark schemes in terms of the network sum transmission rate. Additionally, our proposed scheme exhibits faster convergence compared to the conventional reinforcement learning. Overall, our proposed framework can enrich intelligent resource allocation frameworks and pave the way for realizing real-time resource allocation.

Source: IEEE Transactions on Communications

DOI: 10.1109/tcomm.2025.3547722

Authors: Rui Ding, Fuhui Zhou, Qihui Wu, Derrick Wing Kwan Ng, Kai-Kit Wong, Naofal Al-Dhahir

Institutions: Nanjing University of Aeronautics and Astronautics, UNSW Sydney, University College London, The University of Texas at Dallas

Language: English

Topic: UAV Applications and Optimization

Subfield: Aerospace Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: green

Funders: National Natural Science Foundation of China, National Key Research and Development Program of China