Abstract
Peer-to-peer (P2P) video streaming is an emerging streaming service, which can support large-scale services with a lower infrastructure cost compared to traditional client-server structures. The key of a successful video streaming system lies in the video quality perceived by users. However, one of the major challenges to P2P video streaming services is the visual quality degradation due to packet loss. Because peer dynamics lead to more packet loss opportunities for peers, and the packet loss of a peer will propagate to its neighboring peers through inter-peer transmissions. Such packet loss can seriously damage the reconstructed video quality. To address the packet loss problem, we propose efficient error protection mechanisms for reliable P2P streaming systems. First, we propose a sender-driven peer selection scheme involving estimation of packet loss propagation, evaluation of peers’ contributions, and peer selection based on child-peers’ contributions, for non-scalable streaming over a mesh-based P2P video streaming systems. The proposed packet loss propagation model takes into account the link packet drop rate, peer dynamics, and forward error correction (FEC) protection to characterize the heterogeneous packet loss behavior of individual substreams transmitted over a mesh network. The evaluation of candidate peers’ contributions is modeled through Markov random fields (MRFs) to significantly reduce complexity. We then propose a hybrid sender/receiver-driven error protection scheme to transmit scalable video packets over packet-lossy peer-to-peer networks. A joint source-channel coding (JSCC) scheme based on receiver-driven subscription is proposed to minimize the distortion of the videos received by child-peers under an estimated system uplink capacity by selecting the appropriate amount of source and channel coding packets. Because the bandwidth for inter-peer transmissions may fluctuate largely due to peer dynamics, in our method peers estimate an average system uplink capacity through consensus propagation to avoid the oscillatory allocations of the JSCC scheme. To efficiently utilize the uplink bandwidth of peers, parent-peers utilize sender-driven contribution-guided peer selection to reject the low-distortion contribution subscriptions from candidate child-peers. Simulation results demonstrate that our methods can significantly mitigate packet loss and therefore increase the visual quality, compared to other state-of-the-art schemes.