Multi-client Dynamic Adaptive Video Streaming Control Over Named Data Networking

With the rapid development of communication network and intelligent terminal equipment,video service has become the main source of traffic in the network,and the issue of network congestion is increasingly serious.How to ensure the quality of video service in a complex network environment has caused extensive research.Dynamic Adaptive video Streaming(DAS)is a promising solution.In addition,the Named Data Networking(NDN)can effectively improve the quality of video transmission service due to the advantages of cache in the network,interest package aggregation,multi-path forwarding and data security.The researchers have shown that the performance of DAS over the NDN is superior to that of traditional TCP/IP networks.Given the current network environment,it is increasingly common for users to share limited bandwidth resources.Though some studies have focused on DAS over NDN,they are suboptimal in multi-client scenarios because of changing network topol-ogy,unfair network resource allocation and heterogeneous distribution of in-network cache.Therefore,this dissertation studies the client-driven DAS bitrate control tech-nology.The main work is summarized as follows:1.We construct the DAS architecture in NDN,model the multi-client DAS over NDN problem and design client-driven bitrate adaptive algorithm which needn't the communication between clients or change to routers.However,the clients have no ideal about the network status,and cannot directly obtain the optimal solution,only through a single client rate selection optimization,to improve the quality of experience(QoE)of the overall users.2.The bitrate control algorithm based on utility function(UB)is proposed.When making a decision of bitrate to request,the clients fully consider the benefits and costs,which can reduce aggressiveness of client and improve the fairness of bandwidth shar-ing.Compared with the traditional RB and BB algorithm,UB can obtain a certain promotion in performance.However,the algorithm is single client based in essence and the improvement is limited.3.We model the multi-client DAS over NDN as a multi-stage Bayesian game,the client and remaining part of the network are regarded as the two players in the game and design a reasonable utility function pair.We design a game based adaptive bitrate decision algorithm(GB)for multi-client by solving Bayesian Nash Equilibrium(BNE)problem.The performance is better than the RB and BB algorithm and can achieve fair bandwidth sharing and aggregation of requested bitrate among clients which can make the most of the cache in NDN.4.The multi-client adaptive video streaming control system over NDN is devel-oped and deployed on actual platform to compare and analyze the performance of every algorithm.Furthermore,detailed experiments are carried out on the simulation platform based on ndnSIM,and the performance of algorithms is analyzed at length.