Research On Online Multicast Coded Caching

With the development of network infrastructure and the popularity of mobile devices,online video traffic grew rapidly and thus accounted for more and more proportion in network traffic.Online video is time-varying,which means that the peak network traffic is much higher than the off peak network traffic,so to meet the peak of the user requests,will cause a lot of bandwidth waste in the off peak.Coded caching places some files into user's local storage at off peak,then when the user requests arrive at peak,those data cached do not need send again,which reduce peak network traffic.Besides,the server uses the cache among users to create multicast opportunities,which further reduces the peak network traffic.This thesis studies the online coded caching problem mainly.The online coded caching based on the coded caching,takes cache update process into account to ensure the effectiveness of the user cache.The existing online coded caching schemes are all based on a decentralized scheme,while the centralized scheme has a better performance.Therefore,the application of the centralized scheme to the online scene shows that the performance gain is obtained when the number of active users do not change.Moreover,this thesis also analyzes the situation of the decrease of active users,and shows that it is still able to meet the user requests,but the performance is not optimal.The existing online coded caching scheme selects a file randomly from the server to do replacement when updating the server file,and then update the user cache by keeping the user cache and server files consistent.When the server file popularity is non-uniform distribution,the way of random replacement may lead to replace the file of high popularity out,thereby reducing efficiency of the cache.This thesis uses the least recently sent eviction rule to optimize the existing scheme,that is,replace the least recently sent server file with the new file,which can guarantee the maximum cache efficiency.The result shows that the optimized scheme is superior to the original scheme under non-uniform distribution.Then on the basis of the optimized scheme,the online grouping is investigated,where the high popularity group are allocated to more cache space,so that users can obtain better cache efficiency.The result shows that the online grouping scheme outperforms the optimized scheme.