Recent Advances in Peer-to-Peer Media Streaming Systems

ABSTRACT

Recently, there is great interest in using the peer-to-peer (P2P) network in media streaming. A great number of P2P media streaming systems have been developed. In this paper, we first give a brief survey on some key techniques and algorithms in the field of P2P streaming research. We also analyze the market view of P2P streaming media service, and give a brief descrip-tion about the current mainstream P2P streaming systems deployed in China.

Key words: peer-to-peer, media streaming, the Internet, application-layer multicast

I. INTRODUCTION

The rapid development of the Internet has changed the conven-tional ways that people access and consume information. Besides sending and receiving e-mails, browsing web pages, and downloading data files, people also hope to call telephone, watch movie and TV, and conduct other entertainments via the same Internet. The ideal objective is that anyone can access anything (contents) from anywhere at any time. It is commonly conceived that the next generation Internet should be a multi-media communication network based on the core of IP protocol. Besides traditional data services, other multimedia contents such as voice, image, and video, would also be delivered over the same IP network, among which the streaming media service will play a very more important role.

Streaming media enables real-time and continuous delivery of video and audio data in a fashion of “flow”, i.e., once the sender begins to transmit, the receiver can start playback almost at the same time while it is receiving media data from the sender, instead of waiting for the entire media file to be ready in the local storage. Unlike normal data file, a streaming media file is huge, thus requires high channel bandwidth. Moreover, streaming media also carries stringent demand in the timing of packet delivery. The large size of the streaming media as well as its delivery timing requirement causes a streaming media server to be expensive to set up and run.

In traditional client/server-based media streaming systems, all clients access the same server resource. In this scenario, on the one hand, the processing power, storage capacity, and I/O throughput of the server may become the bottleneck; on the other hand, large number of long-distance network connections may also lead to traffic congestion, thus cannot afford better quality of service (QoS) comparable with that of other tradi-tional Internet services, such as WWW and FTP, and cannot meet the performance requirements of large-scale real-time media streaming applications, especially in the aspects of scalability, adaptability, fault-tolerance and robustness.

To address these problems, recently researchers have pro-posed many solutions, such as IP multicast and CDN (content delivery network). However, both of them need supports from special hardware. For IP multicast network, large-scale multicast-capable routers must be redeployed in the Internet. For content delivery network, a large number of CDN servers should be placed at the network edge, close to any receiver, and cooperate with each other to distribute multimedia data. The costs of infrastructure setup and administration are expensive, and cannot resolve the problems fundamentally.

In recent years, Peer-to-Peer (P2P) networking technology has gained tremendous attention from both academy and industry. In a P2P system, peers communicate directly with each other for the sharing and exchange of data as well as other resources such as storage and CPU capacity, each peer acts both as a client who consumes resources from other peers, and also as a server who provides service for others. P2P systems can benefit from their following characteristics: adaptation, self-organization, load-balancing, fault-tolerance, availability through massive replication, and the ability to pool together and harness large amounts of resources. For example, file-sharing P2P systems distribute the main cost of sharing data - bandwidth and storage - across all the peers in the network, thereby allowing them to scale without the need for powerful and expensive servers.

P2P systems are originally applied in network file sharing, and have achieved great success, such as Napster, Gnutella, Emule, and BitTorrent. However, different from general P2P

Recent Advances in Peer-to-Peer Media Streaming Systems

Gao Wen, Huo Longshe, Fu Qiang

Institute of Digital Media, Peking University, Beijing, China

file sharing, P2P media streaming poses more stringent timing and resource requirements for real-time media data transmis-sion and rendering, therefore it is needed to provide more restricted functions in the respects of resource management, scheduling, and control.

Various P2P media streaming systems have been proposed and developed recently. Even in China, nowadays there are about more than a dozen of P2P streaming applications de-ployed in the Internet. In this paper, we first give a brief survey on some key research issues and algorithms of P2P streaming systems, and then analyze and summarize the current status and development trend of P2P streaming market in China.

II. RESEARCH PROGRESS

OF P2P MEDIA STREAMING

A simple and straightforward way of P2P streaming implemen-tation is to use the technique of application-layer multicast (ALM). With ALM, all peer nodes are self-organized into a logical overlay tree over the existing IP network and the stream-ing data are distributed along the overlay tree. The cost of providing bandwidth is shared among the peer nodes, reducing the burden of the media server. In application-layer multicast, data packets are replicated and forwarded at end hosts, instead of at routers inside the network. Compared with IP multicast, application-layer multicast has several advantages. On the one hand, since there is no need for supports from routers, it can be deployed gradually based on the current Internet infrastructure; on the other hand, application-layer multicast is more flexible than IP multicast, and can adapt different distribution demands of various upper level applications.

Thus, how to construct and maintain an efficient ALM-based overlay network has became one of the key problems of P2P streaming research. To address this problem, mainly three questions should be answered. The first relates to the P2P network architecture, i.e., what topologies should the overlay network be constructed? The second concerns routing and scheduling of media data, i.e., once the overlay topology is determined, how to find and select appropriate upstream peers from which the current peer receives the needed media data? The third is membership management, i.e., how to manage and adapt the unpredictable behaviors of peer joining and departure? Recently, several P2P streaming systems and algorithms have been proposed to address the above issues. From the view of network topology, current systems can be classified into three categories approximately: tree-based topology, forest-based (multi-tree) topology, and mesh topology. In the following we give a brief summarization of P2P streaming techniques accord-ing to this classification.

2.1 Tree-based topology

The typical model of tree-based P2P streaming system is PeerCast [1]. In PeerCast, nodes are organized as a single multicast

tree, where the parent provide service only directly to its sons. The node joining and departure strategies used in PeerCast are simple. For node joining, a new node n first request services from the root node S. If the S has enough resources, it provides service for n directly; otherwise, S redirects the request of n to one of its sons. The son then repeats this process, until the parent of n is found. Since each node only maintains the information of its parent and sons, unbalanced tree may be constructed. Generally, there exist four route selection strategies in PeerCast: random selection, round-robin selection, smart selection accord-ing to physical placement, and smart selection according to bandwidth. To achieve a balanced multicast tree, custom routing policy should be chosen carefully for individual peer node. ZIGZAG [2] is another tree-based P2P streaming system which can construct more balanced multicast tree. ZIGZAG organizes receivers into a hierarchy of bounded-size clusters and builds the multicast tree based on that. The connectivity of this tree is enforced by a set of rules, which guarantees that the tree always has a height O(log

k

N) and a node degree O(k2), where N is the number of receivers and k is a constant. Furthermore, the effects of network dynamics such as unpredictable receiver behaviors are handled gracefully without violating the rules. This is achieved

requiring a worst-case control overhead of O(log

k

N) for the worst receiver and O(k) for an average receiver.

Other tree-based P2P streaming systems also include NICE [3], Overcast [4], and Bayeux [5].

2.2 Forest-based topology

Conventional tree-based multicast is inherently not well matched to a cooperative environment. The reason is that in any multicast tree, the burden of duplicating and forwarding multicast traffic is carried by the small subset of the peers that are interior nodes in the tree. Most of the peers are leaf nodes and contribute no resources. This conflicts with the expectation that all peers should share the forwarding load.

To address this problem, forest-based architecture is beneficial, which constructs a forest of multicast trees that distributes the forwarding load subject to the bandwidth con-straints of the participating nodes in a decentralized, scalable, efficient and self-organizing manner. A typical model of forest-based P2P streaming system is SplitStream [6]. The key idea of SplitStream is to split the original media data into several stripes, and multicast each stripe using a separate tree. Peers join as many trees as there are stripes they wish to receive and they specify an upper bound on the number of stripes that they are willing to forward. The challenge is to construct this forest of multicast trees such that an interior node in one tree is a leaf node in all the remaining trees and the bandwidth constraints speci-fied by the nodes are satisfied. This ensures that the forwarding load can be spread across all participating peers. For example, if all nodes wish to receive k stripes and they are willing to

forward k stripes, SplitStream will construct a forest such that the forwarding load is evenly balanced across all nodes while achieving low delay and link stress across the system. Striping across multiple trees also increases the resilience to node failures. SplitStream offers improved robustness to node failure and sudden node departures like other systems that exploit path diversity in overlays. SplitStream ensures that the vast majority of nodes are interior nodes in only one tree. Therefore, the failure of a single node causes the temporary loss of at most one of the stripes (on average). With appropriate data encodings, applications can mask or mitigate the effects of node failures even while the affected tree is being repaired. Besides SplitStream, there are many other forest-based systems. Examples include building mesh-based tree (Narada and its extensions [7], and Bullet [8]), leveraging layered coding (PALS [9]), and multiple description coding (CoopNet [10]).

2.3 Mesh topology

In conventional tree-based P2P streaming architectures, at the same time a peer can only receive data from a single upstream sender. Due to the dynamics and heterogeneity of network bandwidths, a single peer sender may not be able to contribute full streaming bandwidth to a peer receiver. This may cause serious performance problems for media decoding and rendering, since the received media frames in some end users may be incomplete. In forest-based systems, each peer can join many different multicast trees, and receive data from different upstream senders. However, for a given stripe of a media stream, a peer can only receive the data of this stripe from a single sender, thus results in the same problem like the case of single tree.

Multi-sender scheme is more efficient to overcome these problems. In this scheme, at the same time a peer can select and receive data from a different set of senders, each contributing a portion of the streaming bandwidth. In addition, different from the multi-tree systems, the sender set members may change dynamically, due to their unpredictable online/offline status changes, and the time-variable bandwidth and packet-loss rate of the Internet. Since the data flow has not a fixed pattern, every peer can send and also receive data from each other, thus the topology of data plane likes mesh. The main challenges of mesh topology are how to select the proper set of senders and how to cooperate and schedule the data sending of different senders. Examples of mesh-based multi-sender P2P streaming system include CollectCast[11], GnuStream[12], and DONet (CoolStreaming) [13].

CollectCast puts its emphasis mainly on the judicious selec-tion of senders, constant monitoring of sender/network status, and timely switching of senders when the sender or network fails or seriously degrades. CollectCast operates entirely at the appli-cation level but infers and exploits properties (topology and performance) of the underlying network. Each CollectCast session involves two sets of senders: the standby senders and the active senders. Members of the two sets may change dynamically during the session. The major properties of CollectCast include the following: (1) it infers and leverages the underlying network topology and performance information for the selection of senders. This is based on a novel application of several network performance inference techniques; (2) it monitors the status of peers and connections and reacts to peer/connection failure or degradation with low overhead; (3) it dynamically switches active senders and standby senders, so that the collective network performance out of the active senders remains satisfactory. GnuStream is a receiver-driven P2P streaming system which is built on top of Gnutella. It features multi-sender bandwidth aggregation, adaptive buffer control, peer failure or degradation detection and streaming quality maintenance. GnuStream is aware of the dynamics and heterogeneity of P2P networks, and leverages the aggregated streaming capacity of individual peer senders to achieve full streaming quality. GnuStream also per-forms self-monitoring and adjustment in the presence of peer failure and bandwidth degradation.

Recently, DONet implemented a multi-sender model by introducing a simpler and straightforward data-driven design, which does not maintain an even more complex structure. The core of DONet is the data-centric design of streaming overlay, and the Gossip-based data schedule and distribution algorithm. In the data-centric design of DONet, a node always forwards data to others that are expecting the data, with no prescribed roles like father/child, internal/external, and upstreaming/ downstreaming, etc. In other words, it is the availability of data that guides the flow directions, while not a specific overlay structure that restricts the flow directions. This data-centric design is suitable for overlay with high dynamic nodes. Gossip algorithms have recently become popular solutions to multicast message dissemination in P2P systems [14]. In a typical gossip algorithm, a node sends a newly generated message to a set of randomly selected nodes; these nodes do similarly in the next round, and so do other nodes until the message is spread to all. The random choice of gossip targets achieves resilience to random failures and enables decentral-ized operations. Similar to the related work [15], DONet employs a gossiping protocol membership management. The data sched-ule and distribution method used in DONet is also partially motivated by the gossip concept. It uses a smart partner selection algorithm and a low-overhead scheduling algorithm to intelligently pull data from multiple partners, which greatly reduces redundancy. Experiments show that, compared with a tree-based overlay, DONet can achieve much more continuous streaming with comparable delay.

III. P2P STREAMING IN CHINA

Since the first practical P2P streaming media system was born, P2P streaming service has experienced a significant growth in

China, especially in the year 2005 and 2006. According to a market report [16], over more than 12,000,000 Internet users have accessed P2P streaming service or downloaded P2P streaming software in China. It is predicted that by the end of the year 2006, this number can take a growth to above 25,000, 000. Facing such a large pre-profitable market, till now there are at least over 15 organizations that are providing P2P or likely streaming services. With the most representative, PPlive, PPstream, Mysee, ROX and UUsee have taken over 80% of the current market share. In the rest of this section, we will analyze the market view of P2P streaming media service, and then give a brief introduction to the current mainstream P2P media streaming systems deployed in China.

There are three reasons which cause P2P media streaming service so popular in China in recent years. Firstly, thanks to the rapid advance of audio and video compression technologies, users can easily have access to streaming media in a very low bit rate. More and more multimedia productions, TV clips, and movies are full of the whole Internet. This makes the P2P streaming service providers easier to get enough media sources for service than before. With the various and abundant supply of media contents, service providers can attract more and more clients. The larger the client number, the easier to make test of software and services. Secondly, compared with the traditional way of watching video from the Internet, such as VOD, users can get more satisfied quality of service in current bandwidth-limited network environment. Finally, by the growth of users’network access bandwidth, they demand on more luxury experience, not simply on text and pictures, but more on fluent and high-definition videos. Users’ trend makes a large roomage for P2P streaming service to grow.

Although P2P streaming service has achieved a considerable user experience and definitely it would have a bright future, there are still several issues need to pay attention to. First, current service providers have not found any distinct business models yet. Currently, almost all P2P solution vendors are providing TV program/movie broadcastings free of charge. Obviously, it is not practical for the service provider to charge the users in the time of promoting the service. In the starting period, developing user numbers and gaining subscribers are the key points but earning profits. Second, P2P streaming service providers should face the challenge of copyright. As we’ve just mentioned, some P2P vendors provide TV/movie broadcasting using third party contents without checking their legal status. For long term development, service providers must make cooperation with content providers to make a twin win. Thirdly P2P streaming service providers must face the sur-veillance from the Internet service providers (ISPs) and govern-mental authorities. On the one hand, the purpose of P2P is to maximize the usage of bandwidth resource, however, to the opposite, the bandwidth spewing caused by such applications often makes the ISPs feel intolerable. ISPs usually take rejec-tive actions, such as limiting the application bandwidth or even blocking the application from running on the Internet. However, limiting or blocking is not the most proper way to solve the problem, and the conflicts between the ISPs and P2P streaming service providers will be in existence for a certain while. On the other hand, being regarded as a new media trend on the web, governmental authorities must take surveillance on P2P stream-ing service to guarantee the orderliness of the industry. By the two sides of surveillance, P2P streaming service providers must play the game prudentially.

https://www.wendangku.net/doc/bf15923404.html,, invested by Soft Bank HK, which is acknowl-edged as the number one in terms of subscribers in China, was founded in the early 2005. PPlive has very stable playing quality, and it seldom changes the player’s state to buffering during playing. When watching a new channel, the average waiting time from searching to playing is about 35s to 55s. PPlive provides over 200 channels, categorized by Provincial TV stations, Sports, Cartoon, Entertainment, HK films, Gaming, Movies etc, but very few programs of overseas TV stations. PPlive currently only supports broadcasting, and almost all the program bit rate is between 300kbps~400kbps with media codec like Windows Media Video (.wmv) or Real Media (.rm). Its program timetable is both shown on the website and displayed at the client player. Advertising commercials is supported by the client. Worth to be mentioned, PPlive broadcasted Supergirl Contests in 2005 and it was reported that the concurrent online users hit a record of 500k for the final contest. Though the popular users it has, some contents PPlive provides are lack of copyright, which may be a hidden trouble for its long term development. PPstream, which is founded by two engineers in Sichuan Province, was announced also in the year 2005. Compared with PPlive, PPstream has similar functions but higher connecting speed. Usually when opening a new channel, the average waiting time is about 25s~45s, and its watching fluency is also as good as PPlive. PPstream provides around 90 channels, categorized by Phoenix TV, Wenguang TV, Sports, Entertainment, Movie, TV drama series, Gaming & cartoon, Music and radio channel, and etc. PPstream currently broadcasts Windows Media Video coded QVGA and CIF quality videos with bit rate around 300kbps~440kbps. Its client software supports channel list and timetable shown aside the player, advertising commercials are also supported. It has been reported that PPstream will have cooperation with some ISPs for higher performance, and its market policy seems more steady and long-ranged. Mysee, invested by aurora, which was founded in late 2005, is regarded as a later comer. But Mysee grows quickly in the year 2006. Now, by numbers of media reports, it is very famous on the Internet. Mysee supplies the same video codec like PPstream, but sometimes the connecting speed and playing quality may not be as good as that of PPstream and PPlive. It currently broadcasts around 90 channels which are categorized by news, movie, TV drama series, sports, entertainment, music, information, Cartoon and science. Mysee does not provide a client application player to view programs, all the channels are

viewed in the Microsoft IE browser, channel list and timetable are both displayed on its website. This way may be easy for the service provider to arrange contents that recommended to the user, but lack of user glutinosity. It is reported that Mysee has near one year good cooperation with https://www.wendangku.net/doc/bf15923404.html, and Hunan TV station for video broadcasting. It can be predicted that with preponderance in cooperating with TV stations and ICPs, Mysee would earn a more considerable market share. Roxbeam, used to be called CoolStreaming, is regarded as the first practical P2P streaming software. CoolStreaming was devel-oped in late 2004. It gave a reliable model of P2P streaming. But CoolStreaming was forced to close down due to law suits regarding the content in early 2005. Currently Roxbeam is supported by SoftBank Japan. It not only supplies P2P streaming service, but integrates online community called LeiKe and chatting services into the client software. Users can watch not only broadcasting program but short video clips via the VOD service. Roxbeam tries to provide various video recourses to its user, and its goal is not simply providing a P2P streaming service, but to provide an online video sharing and communication platform. Obviously, Roxbeam has an even grander blueprint, but whether this blueprint can come true is to be proved by the market. UUsee, which is invested by SIG, formed in mid 2005, is also a new power in the P2P streaming service. Having good relationship with CCTV, UUsee has more preponderance than other companions on program copyrights, which can help them much in living broadcast of large-scale activities and programs. UUsee provides about 100 channels on its client player which is categorized by UUsee recommendation, entertainment, sports, movies, TV drama, fashion, cartoon, gaming, science, social news, civil TV stations and etc, channel list and time table are shown friendly on the client player. UUsee also provides thousands of VOD programs on its website, which can effec-tively increase its adhesive ability to the users. By the newest data collection from ACNielsen, during the living broadcast of CCTV’s 2006 Spring Festival Celebration, the UUsee’s user number at the peak time has met the amount of 400,000, which is the largest number from the authority’s report.

By the daily reach statistic from https://www.wendangku.net/doc/bf15923404.html, (http://www. https://www.wendangku.net/doc/bf15923404.html,), in the recent half year, UUsee and PPlive take the first two chairs in the competition, followed with PPstream and Mysee, Roxbeam takes the last. It could be judged that the World Cup in June and Super Girl from May to September contribute more audiences to the Service providers.

Other P2P streaming service providers like QQLive, Pcast, TVants, Poco, 51TV and so on are doing the same contribution to this market. Chance is equal to every competitor, whether they can achieve all depends on the market choice.

IV. CONCLUSION

Recently, P2P streaming has attracted a lot of attentions from both academy and industry. Various P2P media streaming algorithms have been studied, and the systems have been developed. Nowadays about more than a dozen of P2P streaming systems have been deployed in China. In this paper, we first give a brief survey on the progress of P2P streaming research, bring forward some fundamental problems for P2P streaming application development, and review several solu-tions ever proposed to address the problems. Furthermore, we study the factors which can impact the trends of P2P stream-ing market, and make a brief summary for the current P2P streaming market progress in China.

REFERENCES

[1] H. Deshpande, M. Bawa and H. Garcia-Molina, “S tream-ing Live Media over a Peer-to-Peer Network”, Stanford data-base group technical report (2001-20), Aug. 2001

[2] D. A. Tran, K. A. Hua, Tai T. Do, “A Peer-to-Peer Architecture for Media Streaming”, IEEE Journal on Selected Areas in Communications, vol. 22, no. 1, Jan. 2004

[3] S. Banerjee, B. Bhattacharjee, and C. Kommareddy,“S calable application layer multicast”,Proc. ACM SIGCOMM’02, Pittsburgh, PA, Aug. 2002

[4] J. Jannotti, D. Gifford, K. Johnson, M. Kaashoek, and J. O’Toole, “O vercast: Reliable multicasting with an overlay network,” Porc. the Fourth Symposium on Operating Systems Design and Implementation, 2000, pp. 197-212

[5] S. Q. Zhuang, B. Y. Zhao, and A. D. Joseph, “B ayeux: An architecture for scalable and fault-tolerant wide-area data dissemination,” Proc. NOSSDAV’01, New York, Jun. 2001.

[6] M. Castro, P. Druschel, A-M. Kermarrec, A. Nandi, A. Rowstron and A. Singh, “S plitStream: High-bandwidth con-tent distribution in a cooperative environment”, Proc. the International Workshop on Peer-to-Peer Systems, Berkeley, CA, February, 2003.

[7] Y. hua Chu, S. G. Rao, and H. Zhang. A case for end system multicast (keynote address). SIGMETRICS’00: Proc. the 2000 ACM SIGMETRICS international conference on Measurement and modeling of computer systems, pages 1-12, 2000.

[8] D. Kostic, A. Rodriguez, J. Albrecht, and A. Vahdat,“B ullet: high bandwidth data dissemination using an overlay mesh,” Proc. ACM SOSP’0 3, New York, USA, Oct. 2003.

[9] R. Rejaie and A. Ortega, “P ALS: peer to peer adaptive layered streaming,” Proc. NOSSDAV’03, Jun. 2003. [10] V. N. Padmanabhan, H. J. Wang, P. A. Chou, and K. Sripanidkulchai, “D istributing streaming media content us-ing cooperative networking,” Proc. NOSSDAV’02, USA, May 2002.

[11] M. Heffeeda, A. Habib, B. Botev, D. Xu, and B. Bhargava, “P ROMISE: peer-to-peer media streaming using CollectCast,” Proc. ACM Multimedia (MM’03), Berkeley, CA, Nov., 2003.

[12] X. Jiang, Y. Dong, D. Xu, and B. Bhargava. “G nuStream:A P2P media streaming prototype ”. Proc. the 2003IEEE International Conference on Multimedia and Expo (ICME ’03),July 2003.

[13] X. Zhang, J. Liu, B. Li, and T.-S. P. Yum. “C oolstreaming/DONet: A data-driven overlay network for live media streaming ”. Proc. IEEE INFOCOM ’05, 2005.

[14] Zygmunt J. Haas, Joseph Y. Halpern, Li Li. “G ossip-Based Ad Hoc Routing ”. Proc. IEEE INFOCOM 2002

[15] Ayalvadi J. Ganesh, Anne-Marie Kermarrec, and Laurent Massoulie. “P eer-to-Peer Membership Management for Gos-sip-Based Protocols ”. IEEE Transactions on Computers, vol.52, no. 2, Feb. 2003

[16] China P2P Streaming Research Report 2006, iResearch,https://www.wendangku.net/doc/bf15923404.html,

BIOGRAPHIES

Dr. Gao Wen was born in Dalian,Liaoning , in 1956. He received his Ph.D. degrees in computer science from Harbin Institute of Technology,China, in 1988, and in electronics engineering from the University of Tokyo, in 1991 respectively. He joined the faculty of the Harbin Institute of Technology since 1985,

served as lecturer, professor, chairman of department of computer science. He is with Insititute of Computing Technology,Chinese Academy of Sciences, since 1996, serverd as professor,chief scientist, managing director. From 2000 to 2004, he was pointed as Professor and Vice President in Graduate School of Chinese Academy of Sciences, as well as in University of Science and Technology China. From 2006, he became a professor of Peking University. He has published four books and over 300 technical articles in refereed journals and proceedings in the areas of multimedia, data compression, face recognition, sign language recognition and synthesis, image retrieval, and multimodal interface. He has earned three secondary national awards of science and technology achieve-ment in 2000, 2002 and 2003.

Dr. Gao was visiting scientist in the Robotics Institute,Carnegie Mellon University, in 1993, visiting scientist in the MIT AI Lab. in 1995, honorary professor in depatment of

computer science, City University of Hong Kong from 1995 to 1997. He is adjunct professor in University of Science and Technology Hong Kong, from 2004. He severed as the vice chairman of Chinese Association of Image and Graphics, the vice chairman of Chinese Association of Software Industry, the editor for several journals such as the editor-in-chief of Journal of Computer(in Chinese).

Dr. Gao is a leader in some national R&D activities since 1992. He served as the chairman of steering committee for intelligent computing system in 863 Hi-Tech program from 1996 to 2001. He is the head of Chinese delegation to MPEG.He is also the chair of AVS working group which is an entity to make and evaluate the national standard for audio/video coding system.

Dr. Huo Longshe was born in Yangcheng, Shanxi, in 1968. He received his B.S. and M.S. degrees from Xi ’an Jiaotong University,Xi ’an, China, in 1990 and 1993respectively, both in computer science. From 1993 to 2002, He worked in the Designing Institute of Ministry of Posts and Telecommu-nications of China, served as assis-tant engineer, engineer, senior engineer, and department director. He received his Ph.D. degree in Computer science from the Institute of Computing Technology, Chinese Academy of Sciences, in 2006. Now he is a post-doctor researcher in Peking University. Currently his research interests include multimedia communications, P2P media streaming, and multi-view video coding and transmission. He can be reached at lshuo@https://www.wendangku.net/doc/bf15923404.html,

Mr. Fu Qiang was born in Yantai, Shandong, in 1981. He re-ceived his B.S. degree from North-western Polytechnical University in 2004. Now he is pursuing his M.S. degree in the same university,and is a visit student in Peking University. His research interests include multimedia communica-tions and P2P media streaming.He can be reached at qfu@https://www.wendangku.net/doc/bf15923404.html,