基于流量预测的资源分配策略研究

Research on Resource Allocation Strategy Based on Traffic Prediction

在超密集网络中,针对小基站密集部署会导致干扰加剧和资源分配难度加大等问题,提出了一种基于流量预测和着色图的信道资源分配方案。首先,利用深度学习工具实现对网络中未来时隙的流量预测;其次,将不相邻的小基站分配到同一集群以缓减网络中的共道干扰;最后,结合流量预测的结果和着色图为每个小基站分配合适的子信道。通过仿真验证了所提算法的可行性和有效性。结果表明,通过机器学习能够有效预测小基站的流量波动,同时基于流量预测的信道分配方案可以有效降低系统干扰、提升频谱利用率和系统吞吐量。

In an ultra-dense network,a channel resource allocation scheme based on traffic prediction and coloring graph is proposed to solve the problems that the dense deployment of small base stations will lead to increased interference and difficulty in resource allocation.First,use deep learning tools to realize traffic prediction for future time slots in the network;second,assign non-adjacent small base stations to the same cluster to alleviate co-channel interference in the network;finally,combine the results of traffic prediction and the coloring map to assign appropriate sub-channels to each small base station.The feasibility and effectiveness of the proposed algorithm are verified by our simulation.The results show that machine learning can effectively predict the traffic fluctuation of small base stations,and the channel allocation scheme based on traffic prediction can effectively reduce system interference,improve spectrum utilization and system throughput.