Research On Resource Management Driven By User Behavior

With the development of wireless communications,the market share of high traffic services is increasing.Thus,the "user-centric".wireless technology has gradually attracted the attention of the industry.The representative of "user-centric" wireless technology are Device-to-Device(D2D)and Ultra Dense Network(UDN).By extracting new resource forms and integrating terminal equipment and network nodes into the wireless resource management system,D2D and UDN can improve the resource utilization.D2D can bring hop gain,proximity gain and frequency multiplex gain by taking advantage of physical proximity.Similarly,UDN can also bring many gains for network because of physical proximity between access nodes and users.However,facing massive service demands,the wireless resource management system is lack of extensibility and flexibility,especially when dealing with user services which dynamically distribute in the space-time range.Therefore,in order to meet the individual demands of users,save resources and improve utilization of the network,this thesis researches the"user-centric" wireless resource management by taking the user behavior into consideration.This thesis will focus on the problems of link instability in D2D scenarios and frequent handover in UDN scenarios,and then achieve resource management by mining and refining of user characteristics.The main work and innovation of this thesis are as follows:Firstly,this thesis proposes the spectrum and power allocation algorithm based on users' sign-in behavior in the content dissemination network.Based on the users' sign-in behavior data,this thesis predicts the users' trajectory by mining the user association and then predicts the duration of the D2D link.In the content dissemination network,in order to maximize the efficiency of content dissemination,the resource allocation problem is constructed into the evolutionary game based on the predicted duration of D2D links.At the same time,this thesis propose a D2D resource allocation algorithm based on evolutionary game.The simulation results prove that the proposed resource allocation algorithm is better than the contrast algorithms not only in distribution delay,fairness and other user experience indexes,but also in useless throughput and throughput efficiency.And the convergence of simulation results show the feasibility of our proposed algorithm.Secondly,this thesis proposes the access point and energy allocation algorithm based on users' browse behavior in UDN.Based on the users'browse data,this thesis predicts the users' interest by strengthening the user similarity in collaborative filtering algorithm via users' association.Based on the users' interest prediction and the access points' caching,this thesis formulates the access point allocation problem as a many-to-one matching game,and proposes a matching game based access point allocation algorithm.Further,focusing on energy consumption this thesis formulates the transmission loss rate between access points as the adjacency matrix,and solves energy allocation problem based on the shortest path algorithm.Simulation results show that the proposed joint access point allocation algorithm and energy allocation algorithm is superior to the contrast algorithms in terms of throughput,fairness and average update frequency of access point clusters.The proposed algorithm not only provides a good throughput gain in the data plane,but also brings less signaling overhead in the control plane,which is more suitable for the proposed ultra-dense network architecture.