Research On Edge Caching For 5G Networks

Recently,with the development of mobile Internet and the popularity of mobile intelligent terminals,the data services subscribed by mobile users have gradually shifted from ”connectioncentric” communications,e.g.,phone calls and text messages,to ”content-centric” communications,e.g.,multimedia file sharing and video streaming,and mobile data traffic is undergoing an unprecedented worldwide growth,posing a huge challenge to wireless networks.To deal with these challenges,wireless edge caching has aroused more and more attention of researchers.The basic idea of wireless edge caching is to cache popular contents at base stations and user devices during off-peak hours,so that mobile users can get the required contents from local base stations or proximal user devices directly,instead of utilizing backhaul links.Such local acccess can not only reduce the repeated transmission of popular contents and alleviate the heavy burden on the backhaul links,but also reduce the transmission delay and improve users' quality of experience(QoE).One of the key problems of wireless edge caching is to determin how to effectively design cache placement strategies in different scenarios in wireless networks.Surrounding this issue,this thesis studies the cache placement strategies at small cell base stations(SBSs)in multicast-aided heterogeneous cellular networks(HCNs),ultra dense networks(UDNs)and wireless caching networks with user mobility.The main work and specific contributions of this thesis are listed as follows.Firstly,this thesis interprets the background of this research,and reviews current research at home and abroad.Then,a brief introduction is given to the essential problems and main characteristics of wireless edge caching.Moreover,how to apply these characteristics to the design of cache content placement and cache content update is presented.This section establishes the foundation for follow-up study.Secondly,a distributed cache placement and user association scheme is studied in multicastaided heterogeneous networks.The thesis takes advantages of multicast scheduling and cooperative caching to minimize the backhaul cost while ensuring the quality of service(QoS)of all the users.The problem is formulated to jointly optimize user association and cache placement strategy.And the problem can be decomposed into two subproblems,i.e.,optimizing user association policy given the cache placement and optimizing cache placement given the user association policy.Then,the solutions for the two subproblems are provided,respectively.At last,a multicast and cooperative caching algorithm is proposed to jointly optimize the user association and cache placement strategy iteratively.Simulation results show that the proposed scheme can reduce not only the backhaul cost,but also the power consumption.Thirdly,mobility-aware caching in wireless networks is studied.In order to deal with more practical application scenarios,user mobility is exploited to improve existing static caching schemes.In this thesis,user mobility is modelled using a contact model.And then the mobilityaware caching placement problem is formulated to maximize the cache hit ratio.It can be proved that the optimal problem is also a monotone submodular maximization problem with a matroid constraint.To solve it,a greedy algorithm which can provide a near-optimal caching placement Strategy is proposed.Simulation results show that the proposed scheme significantly outperforms random caching scheme and popular caching scheme in terms of cache hit ratio.In addition,Simulation results show that SBSs should cache the most popular files when users move at a relatively low speed,while SBSs should cache less popular files to avoid duplication when users move at a relatively high speed.Finally,a cache placement startegy based on user clustering in UDN is studied.Considering that the existing schemes can not balance between system performance and computational complexity in UDN,a novel cache placement algorithm is proposed based on user clustering.The thesis first formulate the caching strategy for maximizing the number of users served by the SBSs.Then,a spectral clustering algorithm is applied to group users into clusters based on their file requests and physical locations.According to the clustering results,SBSs can be divided into different groups.Aftering grouping,SBSs can only serve users of the class to which they belong.Meanwhile,the cache placement strategy of each class can be designed independently without interference,so that a greedy caching algorithm is proposed to solve optimize cache placement strategy of each class.Simulation results show that the proposed scheme can achieve a good compromise between the computation complexity and system performance.