Monitoring of mesh-pull P2P live media streaming system with random walk based sampling

Mesh-pull based P2P live media streaming system has gained a lot of interests and successes, and collecting key metrics in time from millions of peers to monitor the system level quality is very important for its successful commercial deployment. But the real-time monitoring still faces great technology obstacles, especially considering the dynamic and large-scale scenario which is very common in p2p environment. In this article, we present the novel monitoring method of P2P live media streaming system based on random walks which sample the peers uniformly. The source node initializes many parallel independent random walks periodically to collect the key parameters overall the system and conduct the service quality estimation passively based on the random walk results. The accuracy of the monitoring is influenced by many factors like the number of parallel random walks, the hops of each random walk, the period of sampling and etc. We show the influence of different parameters and verify the effects and efficiency of the method by simulation at last.