A method for uplink power control in cellular networks based on a graph attention-conditional generation diffusion model

By constructing a graph attention-conditional generation diffusion model, the uplink power control of a single-cell cellular network is optimized, solving the deployment problem of traditional methods in highly dynamic communication networks and achieving energy efficiency improvement with low complexity and high generalization ability.

CN121037958BActive Publication Date: 2026-05-26DALIAN MARITIME UNIVERSITY
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202511194023.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-08-25
Publication Date
2026-05-26
Estimated Expiration
2045-08-25

AI Technical Summary

Technical Problem

Traditional optimization methods are difficult to deploy in the context of mobile communication networks with high timeliness and high dynamic change rate. Furthermore, existing AI methods have limited generalization ability in highly dynamic and uncertain communication environments and lack edge computing capabilities, thus requiring lightweight and efficient AI algorithms.

Method used

A graph attention-conditional generation and diffusion model is adopted. By constructing a graph structure dataset of a single-cell cellular network, the graph attention-conditional generation and diffusion model is designed. By leveraging the strong data structure learning ability of GNN and the strong generalization ability of DM, the uplink power control of users is optimized, thereby improving energy efficiency.

Benefits of technology

It reduces computational complexity, has a small model size, and strong generalization ability, which can effectively improve the energy efficiency of uplink transmission for cellular network users and adapt to the power control requirements of large-scale networks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121037958B_ABST
    Figure CN121037958B_ABST
Patent Text Reader

Abstract

This invention discloses a method for controlling uplink power in a single-cell cellular network based on a graph attention-conditional generation and diffusion model. The method optimizes the network energy efficiency of the uplink data transmission model in a single-cell cellular network using the Tinkelbach optimization method. A graph-structured dataset is constructed based on the single-cell cellular network and user equipment (UE). The graph attention-conditional generation and diffusion model is then constructed for uplink power control. This model is trained using the training set and supervised learning target labels from the graph-structured dataset. Based on channel state information and constraint boundaries, the trained graph attention-conditional generation and diffusion model is used to sample the user's uplink transmission power. Based on the sampled user uplink transmission power value, the uplink power of the single-cell cellular network is controlled. The sampled UE transmission power can approximate the upper limit of network energy efficiency to a large extent, thereby ensuring communication quality.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Federated learning for deep neural networks in wireless communication systems

    CN116075834A

  • Non-cellular network power control method based on graph neural network

    CN117560043A