CNIPA.AI
Back to Search
Dossier

一种基于图神经网络的无线网络信道状态预测系统

InventionUnknown
10Claims · 1 independent
§ Ⅰ

Dossier Overview

Inventor

欧阳骞

Filing Date

Jul 27, 2026

Publication Date

Oct 16, 2026

IPC Classification

H04B 17/309 (2015.01)H04B 17/373 (2015.01)H04W 24/06 (2009.01)H04L 1/06 (2006.01)G06N 3/042 (2023.01)

本发明公开了一种基于图神经网络的无线网络信道状态预测系统,包括:信道采集模块,用于生成接入信道观测序列;状态编码模块,用于生成接入信道状态向量;谱映射模块,用于输入引入流形谱稳校核机制的Koopman算子算法;初始构图模块,用于构建初始切换信道图;谱态复核模块,用于生成切换信道图;轨迹生成模块,用于输入改进Graph WaveNet模型,引入信迁谱核序贯联络机制,生成邻接矩阵和候选信道状态轨迹;轨迹整序模块,用于生成信道状态轨迹。本发明提高了无线网络切换场景下信道状态预测的连续稳定性和预测精度,降低了复杂动态传播环境中的轨迹跳变率和切换误触发率。

Sign in to read the full text

Claim 1 is shown. Create a free account to read all 10 claims, the full description, all drawings, translations and legal status.

  • All claims
  • Technical field
  • Background
  • Summary
  • Drawing descriptions
  • Detailed description

External Resources