A Hybrid Approach to Assess the Network Awareness of P2P-TV Applications

Abstract

In this work, we develop a general methodology to assess the level of network awareness and friendliness of P2P-TV applications. The methodology is based on a combination of active and passive measurement techniques and can be applied to any P2P-TV system since it is designed to work considering such systems as a black-boxes. As an interesting case study, we then apply this methodology to PPLive, one of the most popular P2P-TV systems nowadays. Focusing on the video content distribution, we consider several per-path and per-peer metrics, investigating which of them mostly biases PPLive download preferences. Furthermore, in order to refine the picture of PPLive peer selection policy, we not only study the impact of different metrics in isolation, but also assess the joint impact of different metrics at the same time. Our main finding is that PPLive seems mainly bandwidth greedy, but does not show any preference toward peer proximity based on RTT delay; at the same time, our results also suggest that this choice alone may provide a nonnegligible level of geographical clustering among peers as a beneficial side effect.