Research On Decentralized Coded Caching In Wireless Networks

With the arrival of the big data and the proliferation of smart mobile devices,the diversified demand of mobile communication services has contributed to dramatic increase in the traffic volume in wireless networks.The strong temporal variability of network traffic distribution leads to low network utilization during off-peak time and network extremely congested at peak time.Coded caching places the parts of content in the caches of user during off-peak time and use the caches of the users to create coded-multicasting opportunities at peak time,which significantly improves network utilization off-peak and reduces the network load at peak time.As an important Coded caching scheme,more and more attention has been paid to decentralized coded caching due to its excellent performance of network transmission rate,flexibility of prefetching and applicability to changeable networks.This thesis focuses on the research of decentralized coded caching,and the main contributions of the thesis are as follows.1.To solve the problem of low efficiency of decoding algorithm in the group decentralized coded caching,a low complexity decoding algorithm based on the original decoding algorithm is proposes in the thesis.The proposed decoding algorithm greatly reduces the number of signals processed at the user by effectively removing redundant messages in the original decoding operation.Then,the feasibility of the proposed decoding algorithm through theoretical analysis is proved in this thesis.Simulation results show that,without any loss of performance,the group decentralized coded caching with the proposed low complexity decoding scheme can effectively reduce the number of messages processed by the user and improve the computational efficiency of the user decoding compared with the original scheme.2.The group decentralized coded caching scheme with coded prefetching for multi-server systems is studied in this thesis.The proposed scheme realizes the distributed storage of files across servers by encoding files with Maximum Distance Separable(MDS)code.By introducing the rate of MDS code as the degree of freedom of the server,the proposed scheme performs a second encoding operation with MDS code on the encoded packets in each server.The users cache a random subset of the re-encoded packets,which realizes the coded prefetching of the files.Then,the relationship between degree of freedom and transmission rate is analyzed.Simulation results show that the proposed scheme can effectively reduce the transmission rate and reduce the performance gap with the centralized scheme by reasonably selecting the degree of freedom of the server to re-encode the encoded packets in the server.3.A new group decentralized coded caching based on MDS code with coded prefetching for relay networks is proposed in this thesis.Based on MDS coding,the proposed scheme jointly designs the cache placement strategy of both relay node and user and the coded delivery strategy of signal.At the same time,the relay node is used to decode and recover the messages that the server has not sent,which effectively avoids the loss of the transmission rate performance of the relay node to the client network and improves the efficiency of user decoding operation.Then,the impacts of the caches of relay node and user on the network transmission rate in the proposed scheme are analyzed.Simulation results show that the total memory at each user and its connected relay nodes is sufficient to store all the files in the library,the server is not required to transmit during the delivery phase and the proposed scheme has the best performance compared the rate performance with other decentralized schemes.