A scheme for efficient peer-to-peer live video streaming over wireless mesh networks

Abstract

Peers in a Peer-to-Peer (P2P) live video streaming system over hybrid wireless mesh networks (WMNs) enjoy high video quality when both random network coding (RNC) and an efficient hybrid routing protocol are employed. Although RNC is the most recently used method of efficient video streaming, it imposes high transmission overhead and decoding computational complexity on the network which reduces the perceived video quality. Besides that, RNC cannot guaranty a non-existence of linear dependency in the generated coefficients matrix. In WMNs, node mobility has not been efficiently addressed by current hybrid routing protocols that increase video distortion which would lead to low video quality. In addition, these protocols cannot efficiently support nodes which operate in infrastructure mode. Therefore, the purpose of this research is to propose a P2P live video streaming scheme which consists of two phases followed by the integration of these two phases known as the third phase to provide high video quality in hybrid WMNs. In the first phase, a novel coefficients matrix generation and inversion method has been proposed to address the mentioned limitations of RNC. In the second phase, the proposed enhanced hybrid routing protocol was used to efficiently route video streams among nodes using the most stable path with low routing overhead. Moreover, this protocol effectively supports mobility and nodes which operate in infrastructure mode by exploiting the advantages of the designed locator service. Results of simulations from the first phase showed that video distortion as the most important performance metric in live video streaming, had improved by 36 percent in comparison with current RNC method which employs the Gauss-Jordan Elimination (RNC-GJE) method in decoding. Other metrics including frame dependency distortion, initial start-up delay and end-to-end delay have also improved using the proposed method. Based on previous studies, although Reactive (DYMO) routing protocol provides better performance than other existing routing protocols in a hybrid WMN, the proposed protocol in the second phase had average improvements in video distortion of l86% for hybrid wireless mesh protocol (HWMP), 49% for Reactive (Dynamic MANET On-Demand-DYMO), 75% for Proactive (Optimized Link State Routing-OLSR), and 60% for Ad-hoc on-demand Distance Vector Spanning-Tree (AODV-ST). Other metrics including end-to-end delay, packet delay variation, routing overhead and number of delivered video frames have also improved using the proposed protocol. Finally, the third phase, an integration of the first two phases has proven to be an efficient scheme for high quality P2P live video streaming over hybrid WMNs. This video streaming scheme had averagely improved video distortion by 41%, frame dependency distortion by 50%, initial start-up delay by 15% and end-to-end delay by 33% in comparison with the average introduced values by three other considered integration cases which are Reactive and RNC-GJE, Reactive and the first phase, the second phase and RNC-GJE.