Using Peer to Peer and Social Networking to support new models for IP based streaming services Ch. Patrikakis, N. Papaoulakis, P. Sipsas, I. Schmidt National Technical University of Athens, {bpatr,npapaoul}@telecom.ntua.gr,
[email protected],
[email protected]
Abstract. In this paper, the use of a platform for distribution of streaming multimedia over IP through and the use of Peer to Peer technologies in combination with Social Networking are presented. A particular use case scenario is given, which can offer alternative ways for Service Providers to support IP based streaming services to end users, featuring the involvement of the latter in the media distribution architecture in return for reduced pricing. To support the above and also meet the requirements for Quality of Service, state of the art encoding techniques are be utilized, which are based on scalable and multiple description coding. The platform is under implementation in the context of the FP7 project SARACEN.
Keywords: Peer to Peer, Multimedia distribution, IP networking
1 Introduction Online video access popularity has been enjoying a tremendous increase in popularity in the past years. According to “Cisco Visual Networking Index: Forecast and Methodology, 2009–2014”, the global online video community will surpass 1 billion users by the end of 2010, while it would take 72 million years to watch the amount of video that will cross global IP networks during calendar year 2014[1]. The expected increase in bandwidth demand for video distribution over the internet (according to the same report, Global Internet video traffic will surpass global peerto-peer (P2P) traffic by the end of 2010) has been giving thrust to research for efficient ways to broadcast video over IP using heterogeneous networks and different types of devices. On the other hand, more and more internet broadcasters adopt state of the art and alternatives low cost technologies for media streaming in order to address this rapidly increasing demand for bandwidth. As regards the models for distribution, two approaches are deployed so far. The first is IPTV [2] which provides high quality in media but is limited to the users of specific ISP networks and the second is WebTV [3] where the content is available
from any local network connected to the Internet but it lacks of quality and stability in media streaming. Both technologies are considered to be the future of the encoding in IP based streaming media. They allow the introduction of new interactive services which were not possible so far. Considering the strengths and weaknesses of both technologies, it is believed that they are more complementary than competitive [4].
2 Combining the advantages of IPTV and WebTV Both WebTV and IPTV have particular advantages as regards the model which they are called to serve. WebTV offers cost free access to the videos, harnessing the capabilities of public internet (however without any quality guarantee), while IPTV can take advantage of the dedicated infrastructure of consumers to convey the videos ensuring high quality and support for bitrates suitable for HD video (however, with the extra cost introduced through the subscription to a service provider). The combination of the advantages of the two approaches could lead to a reduced cost solution capable of supporting quality of service even over the public internet. In the following sections of this paper, we are going to describe such a solution based on a platform for media streaming able to provide high quality in the media distribution following the IPTV model, while the content is accessible from any network, something that WebTV offers to its users. To achieve this, the platform uses the active involvement of users in the media distribution chain through their organization in a peer to peer media distribution network and sharing of their own infrastructure (devices and access to the network) in order to enjoy the services with low cost. The idea is to involve the users in a model borrowed by nature: Reciprocal Altruism, which has been successfully used in P2P file sharing over the widely adopted BitTorrent [5] through the Tit for Tat model deployed. The success of P2P file sharing has led to adoption of the P2P model for multimedia streaming as well. Several platforms supporting P2P streaming already exist (PPLive, PPStream, Coolstreaming, QQLive, SopCast, Feidian, TV Ants) [6] while research is also underway in order to determine the technologies that could render P2P streaming more robust and Quality aware (P2PNext [7], SARACEN [8]).Peer to peer multimedia networks demand few resources in order to be conformed compared to a traditional multimedia network with central servers. Since every node can both download and upload data there is no need to have many central servers or distributed farms of servers in order to support the multimedia distribution. Nodes receive the multimedia content from their neighboring ones and retransmit it to other peer nodes. In this way, every node receives the best possible quality of the multimedia content without overloading the central server. The limited number of central servers means lower budget for the initial cost of the network set up but also lower operational costs. This allows companies which cannot afford great budgets to build their own media streaming networks. The model of distribution proposed in this paper is based on the use of peer to peer networks for the distribution of the media streams. Every user acts as a node of the distribution network and is able to both download and upload media content from and to other nodes. To achieve more stability in distribution and to enhance it, social
networking features are deployed, in order to support the selection of the peers not only in terms of proximity, but also taking into account the context, which in our case are the viewing preferences and the content being accessed each time. Users of the platform can join any social group according to their viewing or content preferences. Starting from this, nodes sharing specific media content are further categorized according to its proximity, since data exchange between neighboring nodes cases less traffic to the rest of the network. In this way, media streaming is enhanced and network traffic is limited so as that every node receives the multimedia content in the best possible quality depending on its internet bandwidth and the available resources at that moment. In order to support the robust transmission of media an uninterrupted experience of users, the deployment of scalable coding techniques can be used to address problems of nodes leaving the P2P network, and network congestion. Both state of the art techniques, Scalable Video coding [9] and Multiple Description Coding [10] can be deployed.
3 Design and implementation of the platform Following, we are going to describe a possible implementation of the platform for offering better Quality of Experience in a personalised media streaming through the integration of scalable media coding techniques, advanced media transport protocols, and P2P technologies with respect to the user privacy. The main architecture is depicted in the following figure. The platform presented here is designed and implemented in the context of the IST FP7 project SARACEN: Socially Aware, collaboRative, scAlable Coding mEdia distributioN. Broadcaster Content
MDC scenario
Group B Content Aggregation
Content Coding
Content Streaming
End user (Peer‐contributor) End users (Peers‐consumers)
Data – metadata streams
User generated content
Complementary MDC streams
User profile repository
Group A
Figure 1: Scalable media distribution in SARACEN The above architecture on media distribution is a common ground among projects dealing with P2P based streaming media distribution. In most of cases, standard encoding for the media streams is used. However, if the same (i.e. H264/AVC)
encoded streams are being sent, the redundancy will be high and options for video rate adaptation to match instantaneous throughput are very limited, such as skipping some B frames which create noticeable motion jitter. Alternatively, it is possible to generate multiple descriptions from H264/AVC video streams and send a different description from each path to reduce overall transmission redundancy. But even in this case, options for rate adaptation of each stream will still be limited. To overcome the above limitations, streaming of scalable media can be adopted. Transmission of SVC bit streams over each path allows efficient adaptation of the video rate to available network throughput, using congestion control protocols such as DCCP or UDP or TCP-variants like the TFRC. Clearly, sending Multiple Descriptions generated from an SVC stream such that each description itself is scalable should yield the best results in terms of overall redundancy and efficient adaptability of streams to avoid congestion related failures in the network. As it is depicted in the previous figure, the streamed media is broken down into several bit streams that are streamed to different nodes belonging to end users, which act as reflectors participating in a peer to peer distribution scheme. The streamed video content can reach each user from multiple paths in order to increase the robustness of the system against single point of failures in the network and bursts of packet losses. There are several options on the type of compressed video bit stream that can be sent along each path. Finally, trying to contribute the most in satisfying the need for Quality of Experience, it also addresses issues of deploying P2P networking schemes over different domains, and media streaming over heterogeneous networks. In this, seamless adaptation and switching between different network technologies, as well as addressing of firewall and NAT issues are issues that will be addressed in the context of the project.
4 Using a hybrid model for offering access to IP multimedia Following the description of the platform given in the previous section, we will proceed with the presentation of the way this platform can be deployed by a service provider in order to offer access to IP multimedia using a hybrid model that combines the advantages of both WebTV and IPTV, as presented earlier. For the needs of the presentation of the use case, let us consider a TV station or ISP (or a joint consortium) that decides to make available the TV media program of the station or broadcast live events in real time over the internet. Meeting the needs for quality assurance and covering a large audience requires very high budgets, and the corresponding agreements and infrastructures that in cases of small companies render the venture impossible. For thi, reason, the SARACEN P2P platform which could provide a corresponding robust mechanism for media distribution built over the infrastructure of the participant consumers is deployed. The deployment of the platform can be achieved through two different methods. The first is by setting up a private streamer on the broadcaster’s premises with an uplink connection with an uplink connection capable of supporting some 10s of MBps (as it will be used for the
distribution of the stream to some ingress points of the media distribution chain, while the P2P will take over for the full distribution to all nodes according to Figure 1. The streaming software will be available from the SARACEN platform and will provide the initial stream to the P2P nodes of the platform. The other solution is with the use of a small number of SARACEN set top boxes that could receive from the UHF the TV program and feed it to the SARACEN platform (thus acting again as ingress points to the P2P distribution network). In order to avoid the copyright issues the SARACEN platform could check if the program that transmitted has the necessary copyright in order to allow P2P sharing otherwise it prevents it. Coming to the end users, apart from the ability to make use of the P2P network for enjoying enhanced quality and robust transmission, a further advantage is introduced: In cases where downlink bandwidth is inadequate for supporting the full bitrate available (i.e. in the case of High Definition content streaming), while alternative access network technologies are available (i.e. WiFi), neighboring nodes can be interconnected and jointly access parts of the stream, exchanging these parts, thus increasing to total download capabilities of each node, surpassing the downlink capabilities over the fixed network access.
5 Acknowledgments The work presented here has been performed in the context of and has been support by the IST FP7 project SARACEN: Socially Aware, collaboRative, scAlable Coding mEdia distributioN. The authors would like to thank all project partners for their support.
6 References [1] Cisco Visual Networking Index: Forecast and Methodology, 2009–2014, White paper, CISCO, June 2, 2010, available on line at http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/n s827/white_paper_c11-481360.pdf [2] TechRepublic, Introduction to IPTV , available online at http://whitepapers.techrepublic.com.com/abstract.aspx?docid=315495&prom o=100511 [3] abouttheweb.com, About WebTV® and MSN® TV, available at http://www.about-the-web.com/shtml/WebTV.shtml [4] DMR The Magazine for Management and Technology, To be continued: IP TV versus Web TV, available athttp://www.detecon-dmr.com/en/article/iptv-versus-web-tv_2007_06_29/page/4 [5] Bram Cohen, “Incentives Build Robustness in BitTorrent”, May 22, 2003, available online at: http://www.bittorrent.org/bittorrentecon.pdf [6] All streaming media.com, Peer to Peer TV, available online at: http://allstreaming-media.com/peer-to-peer-TV/
[7] P2P-Next research project home page, available online at: http://www.p2pnext.org/ [8] SARACEN research project home page,, available online at: http://www.saracen-p2p.eu/ [9] Overview of the Scalable Video Coding Extension of the H.264/AVC Standard Heiko Schwarz, Detlev Marpe, Member, IEEE, and Thomas Wiegand, Member, IEEE, IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, VOL. 17, NO. 9, SEPTEMBER 2007. [10] V. K. Goyal, "Multiple Description Coding: Compression Meets the Network," IEEE Signal Processing Magazine, vol. 18, no. 5, pp. 74–94, Sept. 2001.
