Locality aware P2P overlay architecture for live video streaming

Abstract

Recently, using Peer-to-Peer (P2P) overlay architectures have become a popular approach for live video streaming over the Internet. Since the peers in these systems are geographically distributed, the communication between them imposes huge redundant traffic into the Internet. One solution to address this problem is utilizing the locality aware algorithms to optimize the neighbor selection policy. However, for real-time applications like live video streaming we need to consider the trade-off between localizing the traffic and service quality. In this paper, we propose a novel pure P2P framework through two phases; (1) Local tracker selection, and (2) Overlay mesh construction. We introduce different locality metrics for neighbor selection. This framework adapts itself to the number of participant peers in the network to guarantee scalability. To the best of our knowledge, this is the first work on the locality in live video streaming over P2P networks. Our comprehensive simulation results show that selecting the neighbors by considering locality concepts reduces (85% in average) the cross Autonomous System (AS) border traffics, compared to existing classic locality aware P2P systems. Moreover, the proposed framework decreases end-to-end delay and distortion (50% and 58% in average, respectively).