Study On Layered Media Multicast Based On Network Coding

Currently, with the rapid development and wide spread of multi-user multimedia applications, multicast becomes a more and more important technology for multimedia content distribution. In a multimedia multicast session, receivers may have different receiving bandwidth and need to receive data at different rates, which is called "network heterogeneity". In the single rate multicast, receivers can only receive data at the same rate, which cannot make full use of network bandwidth. However, in layered media multicast, the source encodes the multimedia data into several layers, receivers receive the corresponding layers based on their own receiving bandwidth, and thus can realize different rates to receive data and make full use of network bandwidth.In the store and forward data transmission mode, network intermediate nodes only store and forward the data, without other processing operations. Network coding is a new data transmission mode, which allows network intermediate nodes to encode the received information and forward the encoded information, and then receivers decode the received encoded packets to the original packets of the source. Compared with the store and forward data transmission mode, network coding can significantly improve network performance, such as increasing network throughput, and balancing network load, etc. Recently, many related works indicate that the combination of network coding and layered coding can greatly improve the performance of multimedia multicast applications with network heterogeneity.This dissertation focuses on related issues on network coding based layered media multicast and its main research works and contributions are as follows:(1) Optimal layer rate allocation in network coding based layered media multicastIn layered media multicast, the network throughput is determined by the rate of each layer. This dissertation allocates the rate of each layer to maximize the sum of the receiving rates of all the receivers. Since the problem is a nonlinear integer programming problem, we design an effective heuristic algorithm for layer rate allocation. Firstly, the algorithm decomposes the network graph into subgraphs based on the number of receivers, and then merges these subgraphs into new graphs according to the number of layers. The minimum value of the maximum flows from the source to receivers in each merged new graph is the corresponding layer rate. Simulation results show that the algorithm can increase the network throughput by 15% to 18% and improve the average bandwidth utilization ratio of all the receivers by 6% to 10%, compared with an existing algorithm.(2) Optimal delay control in network coding based layered media multicastSince the increasing stringent bandwidth and delay requirements of multimedia applications over bandwidth-limited networks, to support the quality of service of multimedia applications becomes more and more important in researching multimedia applications. This dissertation studies the effective path bandwidth allocation to ensure that the bandwidth requirements of the receivers are met and that the delay for all receivers to receive their corresponding layers is optimized, in network coding based multicast layered media multicast. This dissertation gives a mathematical model of the problem and proves its NP-hardness, and therefore proposes a polynomial time heuristic algorithm. Simulation results show that the algorithm can decrease the delay of network coding based layered media multicast, while guaranteeing the bandwidth requirements of the receivers.(3) Fair bandwidth allocation in network coding based multi-session layered media multicastThere are many multimedia applications in the current network, and there usually exist multiple multimedia multicast sessions to provide different services to different groups of users. Therefore, how to ensure that all receivers of all sessions share the network bandwidth resources fairly has become an important research issue in multimedia applications. In this dissertation, we study lexicographically fair bandwidth allocation in network coding based multi-session layered media multicast to allow each receiver of each session to share network bandwidth fairly. Firstly, the problem is modeled as an integer linear programming, and then a polynomial time approximation algorithm is designed. The approximation algorithm obtains the approximate optimal solution of the integer linear programming, through a series of linear programming. For each layer, with the greedy approach, i.e., by solving a series of linear programming, the algorithm makes as many receivers of the corresponding sessions of the layer receive the data of the layer as possible. Simulation results show that the algorithm can make each receiver of each session utilize the network bandwidth resources fairly without sacrificing too much throughput, and its bandwidth utilization is efficient.