Research On User Mobility Predict-Ion Mechanism And Caching Strate-Gy In Heterogeneous Networks

With the development of communication technologies and the rapid increase in the number of user devices,the mobile data traffic has exploded on a global scale.At the same time,user demand for high-rate and low-latency communication is also increasing.The heterogeneous networks can significantly improve network performance by deploying low-power small base stations in traditional cellular networks,attracting widespread attention from researchers.In HetNets,predicting and analyz-ing user mobility not only optimizes resource allocation,but also improves the quality of communication services during the user's movement,which has become one of the current research hotspots.User mobility modeling in HetNets requires consideration of the contradiction between prediction accuracy and model complexity.In addition,in the network scenario where edge caching technology is introduced,how to improve the caching efficiency by using the results of mobility prediction also requires in-depth study.The main work of this paper is as follows:A mobility prediction mechanism based on hierarchical idea is proposed.Firstly,the user's moving behavior is analyzed.Based on the average visiting duration in a day,a clustering algorithm for identifying the frequently visited locations is proposed.Then,the user's location predic-tion is divided into two stages,and the coarse-grained and fine-grained predictions are respectively performed to reduce the complexity of prediction.In the coarse-grained prediction phase,the next most likely visited location is predicted based on the second-order Markov chain with fallback.In the fine-grained predic-tion phase,the hidden Markov model is used to predict the precise position of the user from the temporal and spacial dimensions.The simulation results show that compared with the existing prediction methods,the proposed hierarchical mobility predic-tion mechanism has an obvious improvement in the time consumption of prediction,and achieves a compromise between prediction accuracy and complexity.A caching strategy considering different types of video services and user mobility is proposed.For general network videos,in order to increase the caching efficiency,it is considered to store the coded data of video files in the caches of the small base station.By introducing a delayed offloading technique,the caching problem is described as an optimization problem with the goal of minimizing the average transmission cost of the cached contents.An optimized caching strategy is proposed based on the validity period of user requests to maximize the amount of data downloaded from the local caches.For typical streaming videos,to satisfy the user's need to view the video while downloading,consider placing the original video clips in the caches.Based on the prediction results of user trajectories,a caching scheme suitable for streaming videos is proposed.The simulation results show that compared with the video file popularity based caching algorithm,the proposed caching strategy has lower average transmission cost.