摘要
5G网络的广泛部署与业务激增导致网络流量呈现高度复杂性与动态性,传统预测模型难以满足精度需求。文章提出一种融合长短期记忆网络与时间卷积网络的混合深度学习模型,有效捕捉5G流量的长短期依赖与非线性格局。结合预测结果,设计了基于时变业务需求的动态资源切片调度策略、基于流量特征的基站节能智能启停机制以及基于预测的用户QoE保障策略。仿真表明,本方法在多个真实数据集上预测误差较现有方法降低,优化策略在保障服务质量条件下提升频谱效率,降低基站能耗,显著提升5G网络资源利用效率与用户体验。
The widespread deployment of 5G networks and the surge in business have led to highly complex and dynamic network traffic,making it difficult for traditional prediction models to meet accuracy requirements.This article proposes a hybrid deep learning model that combines long short-term memory networks and time convolutional networks to effectively capture the long short-term dependencies and nonlinear patterns of 5G traffic.Based on the predicted results,a dynamic resource slicing scheduling strategy based on time-varying business requirements,an energy-saving intelligent start stop mechanism for base stations based on traffic characteristics,and a user QoE guarantee strategy based on prediction were designed.Simulation results show that this method reduces prediction errors compared to existing methods on multiple real datasets.The optimization strategy improves spectrum efficiency,reduces base station energy consumption,and significantly enhances 5G network resource utilization efficiency and user experience while ensuring service quality.
作者
徐司放
XU Sifang(China Tower Corporation Anshan Branch,Anshan 114000,China)
