Research On Decentralized Coded Caching Technology For Multi-relay Networks

With the explosive growth of mobile data,the emergence of a large number of smart mobile devices put forward higher requirements for existing infrastructure.In the phase of data link transmission,using limited network resources to efficiently transmit requested file and reduce the link load in the peak phase is the key problem faced by coded caching.Thus,in this paper,the coded caching for multi-relay systems decentralized coded caching is studied.Achievable coded caching schemes are proposed for different application scenarios and compared with existing caching schemes.Firstly,the background and research status of coded caching are briefly described.Then,the centralized and decentralized coded caching proposed by Maddah Ali and Niesen are summarized.Based on decentralized coded caching,a multiple subsystem nested(MSN)coded caching scheme is proposed for heterogeneous user cache capacity.Zero-bit padding,and group-based coded caching schemes are introduced to optimize system partition parameters to reduce the link load during transmission.Hierarchical weighted interference cancellation(HWIC)coded caching is proposed for multi-relay coded interference scenarios,which realizes the interference alignment and interference cancellation gains in transmissers and receivers.A hierarchical threshold segmentation(HTS)coded caching scheme is proposed for non-uniform file popularity situations.The least frequently used(LFU)cache and unicast transmission are jointly designed,and file popularity thresholds are proposed and optimized in different scenarios.The research contents and contributions of this thesis are concluded as follows:1.A multiple-subsystem nested(MSN)coded caching scheme is proposed to solve the link load and system complexity of traditional unicast,group-based and zero-bit padding schemes with heterogeneous user cache capacity.In the case of the number of users is greater than that of files,a all-file transmission scheme is proposed with decentralized coded caching to coordinate the request of the file indexes of the relays,reduce the link load during the delivery and the number of transmission times of the server.In addition,the transmission between relays and users adopts a group-based coded caching scheme.Users are grouped according to their requests.Three-segment delivery is adopted to transmit the coded requested subfiles that are not cached by any user,cached exclusively by a single user and other subfiles to explore the potential coding multicast gains within and between groups.In other cases,the zero-bit padding coded caching can achieve the same link gains as the group-based coded caching and reduce the system complexity.The simulation shows that MSN coded caching can achieve 38.5% and 26.5% link gains compared with traditional caching and zero-bit padding scheme in the case of the number of users larger than files,respectively.2.A hierarchical weighted interference cancellation(HWIC)coded caching scheme is proposed to achieve interference alignment and interference cancellation gains in both transmitter and receivers to solve multi-relay interference subfiles and waste of user cache capacities in traditional interference cancellation scheme.The total degree of freedom(Do F)of the system is considered as the evaluation index.It can reflect the effective file size of a single transmission,and the proposed scheme maximizes the Do F of the system with guaranteeing interference cancellation and the link load in the delivery phase.In the case of the number of single interference-free transmission larger than users,the HWIC coded caching further divides the cache of relays and users and introduces orthogonal unicast.Relay and user cache can be fully utilized to obtain additional Do F compared with traditional caching and interference cancellation schemes.Simulation shows that HWIC can obtain Do F gains of about 1.3 times and about 10.6% more than traditional orthogonal unicast and conventional interference cancellation in the case of the number of subfiles in a single interference-free transmission larger than users,respectively.3.A hierarchical threshold segmentation(HTS)coded caching scheme is proposed to reduce link load and improve cache utilization of LFU and traditional coded caching schemes in the peak transmission.Four different situations are discussed in combination with system parameters such as relays and users cache capacities,and the file popularity threshold is proposed and optimized.Decentralized coded delivery,LFU cache and unicast are designed jointly to design different coded cache schemes for popular and non-popular files to effectively reduce link load during delivery phase.Simulation shows that the HTS coded caching can achieve about 15.8% link gains over the transmission links of source and relays,and 36% link gains over the transmission links of relays and users,which grows with the increase of user cache capacity.This thesis mainly studies the decentralized coded caching in multi-relay system.The MSN,HWIC and HTS coded caching schemes are proposed respectively for heterogeneous user cache capacities,multi-relay transmission interference and non-uniform file popularity scenarios.The transmission rates of each link delivery phase are analyzed and derieved theoretically,and proves the proposed caching scheme is better than uncoded and existing coded caching schemes.Therefore,the proposed coded caching schemes can be efficiently applied to the internet of vehicles,5G small cellular networks,and other scenarios.