Peer Selection and Scheduling of H.264 SVC Video over Wireless Networks Konstantinos Birkos, Christos Tselios Department of Electrical and Computer Engineering University of Patras, Greece Email: {kmpirkos,tselios}@ece.upatras.gr Tasos Dagiuklas Department of Telecommunications, Systems and Networks TEI of Mesolonghi Nafpaktos, Greece Email: ntan@teimes.gr Stavros Kotsopoulos Department of Electrical and Computer Engineering University of Patras, Greece Email: kotsop@ece.upatras.gr Abstract—Scalable video enables flexible video delivery over IP networks. The adaptation of the video content to the network con- figuration and the networking conditions is the great advantage of scalable video. In this research work, on-demand H.264 SVC video delivery over wireless multi-hop single-frequency networks is studied. The main objective is to achieve the highest levels of video quality possible while ensuring fairness among competing video flows over a wireless network. With the main barrier being interference, each video receiver has to decide from which peer it will receive a specific layer. The problem of peer selection and layer allocation is formed as a Mixed Integer Linear Program (MILP). A relaxation as a Non-Linear Program (NLP) is also provided. Simulation results based on real input video sequences verify the validity of the theoretical approach. I. I NTRODUCTION Sharing and distribution of multimedia content, either on a real-time or on-demand basis, has emerged as a principal service in modern IP networks. In addition, the expansion of peer-to-peer networks has substituted the traditional client- server model both in file distribution and multimedia delivery. Via peer-to-peer streaming, users participating in an overlay network can share stored video content or collaborate for the distribution of real-time content. By exercising the power of swarm, peer-to-peer networks overcome the problem of the saturation of servers’ uploading bandwidth. Users contribute to the delivery of content by using their own uploading bandwidth. Regardless of the evident advantages of peer-to-peer stream- ing, there are still numerous challenges to be addressed. The unresponsive and bursty nature of the multimedia traffic, the effect of the underlying network topology, the heterogeneity among the participating peers in terms of computational ca- pabilities and uploading/downloading capacities, the dynamic nature of peers and selfishness pose significant issues which have not been thoroughly studied yet. In general, peer-to- peer systems seem surprisingly robust and the theoretical background under their behaviour is an active research field. On the other hand, wireless peer-to-peer networking has ad- ditional challenges bound to the particularities of the wireless medium. The wide adoption of devices with increased wireless networking capabilities along with the peer-to-peer streaming trend have led researchers in the seek of appropriate methods that enhance streaming performance in terms of perceived video quality. Moreover, the SVC extension of the H.264/AVC standard [1] enables the adaptation of a video stream to the limitations imposed by each device and/or the network. SVC provides scalability of the encoded content in terms of data rate (tem- poral scalability), resolution (spatial scalability) and fidelity (quality scalability). A video stream can be encoded into several scalable layers: a base layer and a set of enhancement layers. Therefore, different representations of the same content can be generated to meet the demands of heterogeneous users and receivers. This characteristic makes SVC an ideal coding standard for mobile peer-to-peer video streaming [2]. In the present work, the problem of peer selection and scheduling of SVC quality layers for serving competing video flows over a multi-hop single-frequency wireless network is studied. Given a network topology and a set of receiving peers that request different video sequences that are pre-encoded and stored in other peers in the network, the overall video quality must be maximized. In other words, it has to be decided which sending peers are going to transmit a number of quality layers to each receiving peers. A MILP formulation of the problem and a continuous relaxation as NLP are proposed. Simulation results based on both solutions are presented. The rest of the paper is organized as follows. Section II summarizes the most relevant previous work in the field. In Section III, the system model and the problem statement are presented. In Section IV, a continuous relaxation of the problem is described. Section V contains simulation results. Finally, Section VI concludes the paper. II. PREVIOUS WORK Multisource streaming of SVC video in mobile ad hoc net- works is presented in [3]. The authors employ an unequal error protection scheme based on Raptor codes and forward error correction. Each source transmits a different representation of the video content in a multiple-description manner. The source block size and the symbol size for each layer are chosen to ensure successful decoding provided that a minimum number of symbols have been received. The peer selection mechanism is based on the minimum-hop-count metric. The 978-1-4673-5939-9/13/$31.00 ©2013 IEEE 978-1-4673-5939-9/13/$31.00 ©2013 IEEE 2013 IEEE Wireless Communications and Networking Conference (WCNC): NETWORKS 2013 IEEE Wireless Communications and Networking Conference (WCNC): NETWORKS 1633