Social Television: Enabling Technologies and ... - IEEE Xplore

INVITED PAPER

Social Television: Enabling Technologies and Architectures The concept of TV and other content being everywhere has been accepted in the entertainment world and this paper reviews the recent developments that will create social television experiences beyond the current generation. By Marie-Jose´ Montpetit, Senior Member IEEE , and Muriel Me´ dard, Fellow IEEE

ABSTRACT | In this paper, we review recent networking developments that will help create the next-generation social television experiences. These include revisiting the way networks are created and using social connectivity to drive physical connectivity and network virtualization. Multipath dissemination and reduction of interruptions will provide better quality of experience. Content protection and privacy are also essential to enable social commentary and metadata applications and will be briefly introduced. Examples of potential applications and results of field trials are also included. KEYWORDS | Network coding (NC); network on demand; social networking; social television

I. INTRODUCTION Let us look into the future: After a long day of work, Lisa decides to take a break and watch that TV show that she and her daughter really like. With her daughter in college on the U.S. west coast, Lisa will see the show three hours ahead but wants to make the show as much a common experience as it was last year. To do so she leaves comments and recent related pictures for her daughter to see using her companion application on her smartphone. About halfway through the show she gets a message from her husband to share in his live experience of the Chinese New Year celebrations captured by his smartphone camera. She sends the video to her tablet application so that the main

Manuscript received February 14, 2012; accepted February 21, 2012. Date of publication April 6, 2012; date of current version May 10, 2012. The authors are with the Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, MA 02139 USA (e-mail: [email protected]; [email protected]). Digital Object Identifier: 10.1109/JPROC.2012.2189804

0018-9219/$31.00 Ó 2012 IEEE

screen continues to display her show but during the commercials she switches the screens. At the end of the show she records the rest of her husband’s video on her cloud application for future viewing. Later, her daughter Suzie wants to watch her show but she is in a crowded bus with four of her college friends so they decide to watch it on their own devices; three of them are already subscribing to the TV service with enhanced web experience and the fourth one is allowed to watch after a commercial. At times in the show Suzie sees the messages and pictures her mother left her but her friends can only wonder why she is smiling so much. The future is however very near. The concepts of BTV everywhere[ and Bcontent everywhere[ have readily been accepted in the entertainment and publishing world with devices like smartphones and tablets now being the extensions of the living room and library. The Bsecond screen[ experience is now standard and the use of Facebook or Twitter for content recommendation is common. And the idea of exploiting metadata for social interactivity has also been explored [1]. As seen in Fig. 1, the social filter creates personalized and group-centric viewing experiences that are replacing the traditional approach to television content consumption. But the use case also emphasizes the requirements of socially connected TV: international or time shifted, cloud based, heterogeneous in devices and networks, private and respecting of content protection. In this paper, we review recent developments that will create social television experiences beyond the current generation. We need to revisit the way information is disseminated and how networks are created. We include examples of network coding (NC) as one of the enabler of new network paradigms. As proven by many recent studies and implementations, NC demonstrated improved network Vol. 100, May 13th, 2012 | Proceedings of the IEEE

1395

´ dard: Social Television: Enabling Technologies and Architectures Montpetit and Me

Fig. 1. Using social networking in the TV experience; adapted from [2].

performance for content distribution and protection and for optimizing transmission control protocol (TCP)-based services, the basis for the next-generation of social television. In Section II, we will present architectures using social connectivity to drive physical connectivity and we define how network virtualization beyond the current Bsoftware defined networks[ (SDNs) [3] could be used to create that connectivity Bon-demand.[ Improved dissemination and reduction of interruptions to provide better quality of experience as well as examples of potential applications and results of field trials are the focus of Section III. Content protection and privacy are also essential to enable social commentary and metadata applications; they are presented in Section IV.

II. SOCIAL CONNECT IVITY T O DRIVE PHYSICAL CONNECTIVITY Many social media applications require that the members of the communal experience be connected via some Internet technology using the same network technology or even the same devices. But inherently, social connectivity should not imply physical connectivity, but instead should allow changing the physical connectivity to offer the best quality of experience. In addition, models that until a few years ago seemed to describe the behavior of Internet traffic have become obsolete. Indeed, the Internet is less and less about flows moving from a source to a destination but more and more about information disseminated across a large number of nodes. The information superhighway has evolved into a complex organism of information diffusion. In this context, a re-architecture of the way we consider IP packets becomes necessary and a more diffusive model needs to be defined; one that takes advantage of multipath and combination opportunities. Our need to connect socially should drive our physical connectivity, then if we all want to watch together we 1396

Proceedings of the IEEE | Vol. 100, May 13th, 2012

should be able to download Bthe right network[ to enable it on demand and embody the latest strategies. It would further allow combining and supporting innovation in the academic and application world at the same time. Our answer is to define a Bnetwork as a service[ approach leveraging current cloud computing and Bsoftware as a service[ approaches and enabling radical new technologies based on NC to be deployed efficiently. In fact, futurologist Daniel Burrus had identified Bcloud computing,[ Bon-demand services,[ and Bvirtualization[ as numbers two, three, and four of the 20 technology trends for 2012 [4]. Combining these and pushing these ideas further, a communal viewing group could define its own television network and enable viewing on any screen (as illustrated in Fig. 2). One use case could be for a video device to require a more efficient delivery mechanism in a wireless network, another for a content provider or consumer to request more protection when transmitted over a nonsecure network. A social viewing application would learn of available network services (via service discovery, provisioning, or other) and the policies for requesting (including user authentication and location). It is assumed this capability could eventually become part of the network operating systems but retain its external call functionality. Network performance enhancements like NC could be available to services and application to improve quality of experience especially to better manage content distribution in a web space more and more dedicated to Bwhat[ and less to Bwhere.[ While we are obviously far from this being implemented in a product, the experience of Bwidgets[ and browser plug-ins to allow the specific applications to be deployed or adding features to existing services as well as cloud computing services like those offered by the Amazon Elastic

Fig. 2. neXtream concept from [5].


Computing service points to networking solutions based on service requirements.

II I. NETWORK CODING S TRATEGIES FOR SOCIAL T ELEVISION DELIVERY The traditional television delivery system, based on content being acquired and distributed to a single end device under the control of a single operator is now obsolete. The original content can be combined with ancillary content and extra features that could be inserted anywhere in the network and rendered on an ecosystem of devices over wired and wireless networks alike. Because of the scarcity of resources for video, there are increasing demands for content to be shared locally and to reduce the volume of content management, retransmission of lost segments, and other nonrevenue generating traffic. And NC has proven to be an effective solution for these requirements. NC considers data traffic as algebraic information [6]. The output of a network coder is a linear combination of a number of input packets multiplied by coefficients of a Galois field. It has been shown in an information theoretic manner to reduce the required number of transmissions to complete a file or stream operation over noisy or unreliable networks. Considering that all participants of social TV will not experience the same network conditions, this provides a strategy for success. Finally, since network codes are composable, hence the information is capable of being recombined in the network, they are ideal to communicate socially enhanced video streams without the need to decode before the destination, hence reduce the effects of accumulated delays on the quality of the experience. NC does add complexity to both source and destination nodes but these are quite simple linear operations and NC has been successfully implemented in mobile devices such as phones and MP3 players. The use of random linear network coding (RLNC), where the coding coefficients are chosen randomly over a chosen small Galois field, creates the simplest encoder [7]. The application of NC to real-

world problems is recent but it has already demonstrated its usefulness. One major recent success is the use of NC to improve the performance of web traffic in wireless networks. The analysis of NC with the prevalent protocol of the Internet TCP was in terms of loss rate, round trip, and window sizes [8], [9]. The main aspect of the work is to ensure that losses due to poor channel quality are not interpreted as congestion by TCP. The main element of the protocols is to insert a Bshim[ layer between IP and TCP where datagrams are coded with RLNC and the coded coefficients are added to the header along with the number of packets combined. In addition, in TCP–NC, the acknowledgment (ACK) mechanisms are modified to acknowledge degrees of freedom instead of individual datagrams. It introduced the concept of Bseen[ packets, in which the number of degrees of freedom received is translated to the number of consecutive packets received. Note that NC only prevents random erasures, not congestion. With correlated losses, TCP–NC follows TCP congestion mechanisms and the transmission window is closed. In order to validate these results, trials were conducted around the Boston area with the TCP–NC encoder implemented in Amazon EC3 and the decoder on a laptop computer. The results are shown in Table 1. As can be seen, in very congested areas (Starbucks and downtown) the gains are significant and allow watching video without interruptions and loss of synchronization with social commentary. The success of the TCP–NC opens the way for other theoretical work to move into the application realm. The two most promising are network combining and caching to improve the network and open new architectures to support the creation of the networks presented in Section II. In particular by combining less than 5% of network coded packets from a fourth-generation (4G) network to the packets of a stressed WiFi network the full quality of a 4G session can be achieved and at a much lower cost for both users and operators [10].

Table 1 TCP/NC Trials in the Boston Area

Vol. 100, May 13th, 2012 | Proceedings of the IEEE

1397


IV. SOCIAL CONTENT DISSEMINATION, PROTECTION, AND PRIVACY The scenario presented in the Introduction included the communal viewing of content on individual devices. This was to highlight the problems with the current architecture for such experiences: each device is considered individually with unicast traffic wasting bottleneck resources. Furthermore, device authentication creates nonrevenue generating traffic. Hence, in a social-video-rich wireless environment, getting the information to its destination is not sufficient for commercial success: efficient content dissemination and content ownership need to be protected, hence, the need for novel approaches to content protection in a social viewing perspective. Peer-to-peer mechanisms need to be recognized as a legitimate mechanism for content dissemination [11] while still allowing both protection of commercial content and privacy of social commentary. Digital rights management (DRM) is more about business cases than content protection per se and is usually for a single device and a single user. Heavily encrypted approaches fail to meet the needs of commercially produced content that needs protection when this content is shared among devices and Bfriends[ and gets annotated and enhanced. The commercial content and the annotation may need protection but most likely not the same. One aspect of NC that is yet to be fully exploited is how its algebraic structure inherently protects the encoded content. We propose an approach for distributed content verification without the need to contact a centralized trusted authority. Our techniques build upon our earlier work on constructing encrypted network coded messages [12] in which only the NC coefficients are encoded which in essence protects the whole encoded packet: without knowledge of the coefficients the information cannot be intercepted. The decoding time is very short and significantly faster than traditional decryption and leads to efficient peer-to-peer implementation. And

REFERENCES [1] J.-Y. Hwang, P. Pia i Contesa, H. Holtzman, and M. J. Montpetit, BCommenTV: A time-sensitive social commenting system for audiovisual content,[ in Proc. 9th Annu. IEEE Consumer Commun. Netw. Conf., 2012. [2] R. Martin, A. L. Santos, M. Shafran, H. Holtzman, and M. J. Montpetit, BneXtream: A multi-device, social approach to video content consumption,[ in Proc. 7th Annu. IEEE Consumer Commun. Netw. Conf., 2010, DOI: 10.1109/CCNC.2010. 5421599. [3] B. Lantz, B. Heller, and N. McKeown, BA network in a laptop: Rapid prototyping for software-defined networks,[ presented at the ACM Sigcomm Hotnets Conf., Monterey, CA, Oct. 20–21, 2010.

1398

while peers can help to disseminate NC video content they will not be able to decode it without the right key information (these can be obtained via traditional mechanisms or provisioning). This approach has been implemented successfully on smartphones. But to ensure that the content itself is the original one, we are adding a homomorphic signature scheme that will allow validating the received content even before decoding hence ensuring that the content ownership will be preserved. This work is ongoing.

V. CONCLUSION This paper presented a vision for the use of novel technologies for social television viewing. We believe that to provide the radically new services, combining social connectivity to video streaming and personal content demands redefining the way we currently interact with our network. Social TV, which combines wired and wireless networking, necessitates a comprehensive end-to-end and top-to-bottom strategy. In this view, NC targets the providing of resiliency and efficiency the operators charge for and the flexibility and social connectivity demanded by application developers and consumers alike. The solution to providing social television is not only to use higher capacity networks or better compression mechanisms but also to use the existing networks optimally. The results presented in this paper will apply directly to rising community-based video distribution networks linking ecosystems of smartphones, home gateways, and caches as well as set-top boxes and digital video recorders (DVRs) to efficiently use existing resources and provide the infrastructure for the social connectivity of the future. h

Acknowledgment The authors would like to thank S. Ng of NBC Universal for the invitation to this special issue.

[4] D. Burrus, Technology-Driven Trends for 2012. [Online]. Available: http://www.burrus.com/ trends-for-2012/. [5] R. Martin, A. L. Santos, and M. Shafran, neXtream, MIT MAS 571 class project, 2009. [6] M. Me´dard and A. Sprintson, Eds., Network Coding: Fundamentals and Applications. New York: Academic, 2012. [7] T. Ho, M. Me´dard, J. Shi, M. Effros, and D. R. Karger, BOn randomized network coding,[ in Proc. 41st Annu. Allerton Conf. Commun. Control Comput., 2003. [8] J. K. Sundararajan, D. Shah, M. Me´dard, M. Mitzenmacher, and J. Barros, BNetwork coding meets TCP,[ in Proc. IEEE INFOCOM, Apr. 2009, pp. 280–288.

Proceedings of the IEEE | Vol. 100, May 13th, 2012

[9] M. Kim, M. Me´dard, and J. Barros, BModeling network coded TCP throughput: A simple model and its validation,[ in Proc. ICST/ACM Valuetools, May 2011. [10] A. Kulkarni, M. Heindlmaier1, D. Traskov, M. Me´dard, and M.-J. Montpetit, BNetwork coding with association policies in heterogeneous networks,[ in Proc. NC-Pro, Mar. 2011. [11] M. J. Montpetit, Community Networking: Getting Peer-to-Peer Out of Prison, Jan. 18, 2008. [Online]. Available: cfp.mit.edu. [12] L. Lima, S. Gheorghiu, J. Barros, M. Me´dard, and A. T. Toledo, BSecure network coding for multi-resolution wireless video streaming,[ IEEE J. Sel. Areas Commun., vol. 28, no. 3, pp. 377–388, Apr. 2010.


ABOUT THE AUTHORS ´ Montpetit (Senior Member, IEEE) Marie-Jose received the Ph.D. degree in electrical engineering and computer science from Ecole Polytechnique of Montreal, Montreal, QC, Canada, in 1991. She is a Research Scientist in the Research Laboratory of Electronics, Massachusetts Institute of Technology (MIT), Cambridge, focusing on network coding for video transmission. Her research interests include converged video applications, social and multiscreen media dissemination, and wireless networks. Dr. Montpetit was a recipient of the MIT Technology Review TR10 in 2010.

´ dard (Fellow, IEEE) received the B.S. Muriel Me degrees in electrical engineering and computer science and in mathematics in 1989, the B.S. degree in humanities in 1990, the M.S. degree in electrical engineering in 1991, and the Sc.D. degree in electrical engineering in 1995, all from the Massachusetts Institute of Technology (MIT), Cambridge. She is a Professor in the Electrical Engineering and Computer Science, MIT. Her research interests are in the areas of network coding and reliable communications, particularly for optical and wireless networks. ´dard was named a 2007 Gilbreth Lecturer by the National Prof. Me Academy of Engineering. She is a member of the Board of Governors of the IEEE Information Theory Society.

Vol. 100, May 13th, 2012 | Proceedings of the IEEE

1399