基于聚类算法内容流行度预测的空天地一体化网络缓存方法  

A Space-Air-Ground Integrated Network Caching Method Based on Clustering Algorithm Content Popularity Prediction

作　　者：仇超 王晨阳 QIU Chao;WANG Chenyang(Tianjin University,Tianjin 300072,China;College of Computer Science and Software Engineering,Shenzhen University,Shenzhen 518060,China;Guangdong Laboratory of Artificial Intelligence and Digital Economy(SZ),Shenzhen 518132,China)

机构地区：[1]天津大学,天津300072 [2]深圳大学计算机与软件学院,广东深圳518060 [3]人工智能与数字经济广东省实验室(深圳),广东深圳518132

出　　处：《天地一体化信息网络》2024年第1期40-47,共8页Space-Integrated-Ground Information Networks

基　　金：国家自然科学基金青年项目资助(No.62002260)。

摘　　要：针对空天地一体化网络中恶劣的自然条件可能造成连通异常和通信阻塞的情况,对一种新的空天地一体化网络缓存方法进行研究,即基于聚类的内容缓存方法。基于空天地一体化网络架构,利用提取的内容特征来预测流行度,并用于缓存替换决策。内容的流行度预测是由动态聚类算法实现的,这种方法可以有效地减轻地面核心网的通信压力,从而进一步提升用户体验。使用开放数据集对算法进行评估,实验仿真结果证明了该方法的可行性和有效性。This paper proposed a new space-air-ground integrated network caching method,that was,cluster based content caching method.Based on the space-air-ground integrated network architecture,this method used the extracted content features to predict the popularity and used it for cache replacement decision.Content popularity prediction was realized based on dynamic clustering algorithm.This method could effectively reduce the pressure on the ground backbone network and further improve the quality of user experience.It evaluated the performance of our method using the open Douban movie review dataset.This dataset could be used to simulate user requests.Experimental simulation results showed the feasibility and effectiveness of the method.

关 键 词：空天地一体化网络 内容缓存 流行度预测 聚类算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]