| Recently, network coding based P2P live streaming system turns into a worldwide hotspot for the P2P research community. Contrast to the tradition P2P streaming system, with the help of network coding, simple and efficient transmission protocols can be designed, and the system becomes more robust to the highly dynamic networks and the scalability improves. Based on these advantages, P2P live streaming system with network coding will be promising pattern for the future internet media.Although many researches of the protocol design and the system implementation about network coding based P2P live streaming have been conducted, there are still problems deserving study. For example, due to information equivalence property of the coded blocks, protocols often utilize multi-source push mechanism, however, lack of cooperation control results in large amount of linear dependent redundant data; Performance evaluation of the algorithm in large scale P2P system is another unsolved difficulty, as the development of the prototype systems consumes huge resources and long cycles, while the existing simulator cannot fully support the network coding operations.To handle the above two problems, we propose a source peer selection algorithm for the redundant optimization and illustrate our simulator design proposal for efficiently algorithm evaluation. The details are as follows:We propose selecting active source peers from the source set given by the tracker to reduce the redundant data. Active source peer selection problem can be regarded as a variant of0-1knapsack problem, which is NP hard, such that we design an approximation algorithm to compute the solution. Through formal proofs and simulations, we verify the validity of our algorithm, which can further improve the QoS of the P2P live streaming.To improve the efficiency of algorithm evaluation, we propose a new simulator architecture. This simulator utilizes discrete event driven mechanism as the engine to forward the simulation time, and uses the objects derived from five basic classes, including node, event, protocol, message and operation, to customize simulations for P2P streaming system with diverse protocols. Our simulator owns the features of high level modularization, small quantity of codes, configuration resilience. It can evaluate all kinds of metrics of streaming systems, such as, initial delay and playback fluency, and can also support the network coding operation as well as the network dynamics.In the future, we intends to take up the researches on designing a lightweight cooperation control mechanism, a segment granularity selection algorithm and a mathematic modeling approach for P2P live streaming system with network coding to further improve its QoS. |
