Wiki Paper: Is High Definition a natural DRM?

Information regarding wiki-fied papers hosted here

Title Block / Table of Contents

Title: Is High Definition a natural DRM?

Authors: Andrew Blaich, Aaron Striegel

Embed citation link here

Abstract

DRM, Digital Rights Management, has become prolific. It is used on both physical mediums (including CDs, DVDs, Blu-Ray discs) and digitally distributed content. DRM controls how, when, where and by whom content gets used. However, perfect DRM remains elusive with each variation often being cracked or circumvented by various hacking groups shortly after its release. The use of DRM grew out of the need to protect the distribution of content with file-sharing networks, which grew in size due to the proliferation of broadband internet services. However, for file-sharing to be effective, users need to have sufficient upload bandwidth relative to the memory size of the content being shared. With compressed audio and video content, file-sharing has been relatively successful, but for content that requires a memory footprint a magnitude larger than typically shared before, distribution becomes significantly harder. With the proliferation of high-definition content we argue that DRM in the traditional sense no longer appears necessary. Specifically, we posit that using the natural file size of true high-definition content essentially acts as its own form of DRM due to the extreme asymmetry of broadband speeds and vested economic incentives of ISPs with regards to enforcement.

Introduction

Digital Rights Management, or DRM, is the moniker used for the system or systems that are employed to manage digital multimedia content \cite{drm, drmandprivacy, skeptdrm}. There are numerous variations on DRM and each new distribution channel tends to use its own form. With each iteration of DRM there is the claim that it will be resilient against attacks and circumvention \cite{decodingdrm}, however, this has yet to prove true as is the case most recently with BD+ and AACS (the DRM used on Blu-Ray Discs) \cite{bdcracked, aacscracked}. Creating DRM has become an arms race between a small subset of users that continues to crack and circumvent DRM and the distribution channel used to supply the content that is DRM enabled \cite{halflife}.

With the advent of an increasing amount of high-definition content being distributed digitally as well as in the form of high definition video play-back discs, i.e. Blu-Ray, DRM has been increased to even more pervasive levels. As DRM is continually upgraded users are required to update their video-playback device's firmware \cite{brfirmware}, in order to support newer levels of DRM or else content will not play, or if the user was an early adopter they may need a brand new player or display because their original hardware cannot be upgraded via software \cite{bluraynoplay, hdcpnw}. Systems that are capable of high-definition content playback are essentially full powered computers, requiring large increases in the processing power for both the bit-rate of the content as well as the need to be in a constant protection mode in order to ensure a raw data signal from the content is not user accessible [Gutmann07].

As DRM continues to evolve (software/hardware upgrades) and as the processing power of the players continues to increase because of higher bit-rates in high definition content, we ask if DRM in its current form is even necessary for HD based content? For example, a typical Blu-Ray disc holds 25GB for a single layer and 50GB for a dual-layer disc \cite{brdsp}. This large amount of space is needed in order to supply the proper resolution and bit-rate to keep content in full high definition\footnote{High definition here refers to 1080p quality at a 40Mbps bit-rate.} at the requisite bit-rate (40 Mbps for Blu-Ray \cite{bitrates}). The content size has increased nearly a magnitude over that of a traditional dual-layer DVD (4.7GB for a single layer and 8.54GB for a dual layer disc), and any sort of compression on the signal will typically result in visual artifacts detracting from the HD experience. Additionally, in the United States, the Internet connections are infamously asymmetric for consumers, with the classification of broadband in the US \cite{fcc, fccupgrade} not the same as in other countries such as Japan where broadband is typically 8Mbps. With all of the above factors combining, it is simply not as easy to distribute high-definition video content due to the file sizes and available upload/download bandwidth to the consumer.

The rest of this paper looks at what is currently being shared on file sharing networks. It compares and contrasts the advantages and disadvantages of using High Definition as a natural DRM and how it will operate in the future, as connection speeds and symmetry improve with time. Finally, we examine how endusers can benefit from such improvements.

Background

In order to put into perspective the content available on the Internet, one of the more popular BitTorrent trackers, The Pirate Bay, was searched for Blu-Ray disc movie rips. A quick glance at a subset of the search results, as seen in Table \ref{table-tpb-top10-bluray-sizes}, revealed several Blu-Ray high definition disc rips being tracked. The file sizes averaged around 30GB per movie, with a max size of 42GB observed. As expected, the content is being shared due to the cracking and ripping efforts of individuals and groups scattered throughout the Internet and world. The DRM protection systems used, AACS and BD+ were fully cracked and circumvented, \cite{bdcracked, aacscracked}, allowing these rips to be produced and shared.

Pirate Bay - CD-sized movies (<= 640 MB)	Pirate Bay - Blu-Ray sized movies (30-40 GB)

\begin{figure} 
\centering\includegraphics[scale=0.50]{images/currentdrm.eps} 
\caption{Simplified view of the current DRM model for High Definition Content}\label{fig-current-drm}
\end{figure}

It should be important to note that even though the Blu-Ray discs are being ripped and distributed online, the top 100 movies as listed on The Pirate Bay's tracker are still only DVD rips with an average file size of 700MB (see Table \ref{table-tpb-top10-all} for a subset of the search results). While the small file size can be due to many factors such as trying to fit a movie onto CD or multiple movies onto a DVD, the small file size can also be attributed to the asymmetry of broadband connections. Due to the asymmetry there is a need to distribute content quickly to multiple clients. However, this assumes all users are participating and are actively uploading and downloading \cite{msoftbt} from the swarm. BitTorrent? as a file sharing service relies on users to be both contributers as well as consumers of the tracked content. The protocol suffers when large groups of peers in a swarm are leeching (downloading only) content rather than seeding (uploading) parts that have been successfully download while continuing to download other portions of the requested file. However, even though some users may be seeding, their upload bandwidth is limited, when compared to download speeds, which leads to slower download times of those receiving content from users with limited upload speeds.

The above helps explain why the top files are still measuring approximately 700MB for a feature length movie. The rippers are balancing between quality and size, since larger files will require more people to share longer. Compressing a high-definition video file is possible as it is with DVD quality video. However, HD is more prone to show artifacts after compression \cite{hdcompress}.

As cited by \cite{freeriding}, a torrent has a life-cycle where its popularity dies off leaving fewer and fewer seeders as time increases. Most users it seems, as is the case with the popularity of flash based media sites, including YouTube and Hulu, are content to watch lower resolution videos on their computers, and is easier to be distributed due to the smaller file size. To get a sense of the magnitude involved with sharing full HD content, consider the asymmetric network of today's Internet and as a thought exercise, access speeds envisioned by various Clean Slate architectures (100 Mb/s symmetric). Assuming that the user in a full 100 Mb/s environment could receive the movie at a full 100 Mb/s (ignoring layer 2,3,4 overhead), the effective download speed would be 12.5 MB/s for a 30 GB movie, which translates to a 40 minute download in the best case. In contrast, the current broadband market has little symmetric bandwidth with upload speed often multiplexed heavily between users.

Advantages of using a natural DRM

The advantages of using a natural DRM are obvious, when contrasted against the current model for high-definition DRM as seen in the figure to the right. The numerous levels of DRM checking and re-checking that are necessary for current DRM are not necessary when the natural DRM of high-definition content is employed.

Content providers that provide the full high definition version of their work essentially are giving little incentive for an end-user to share it. With the file sizes for a Blu-Ray disc averaging around 30GB and bandwidth speeds limited, the file size essentially protects the content itself without added help. The main method discussed in this paper for the sharing of content is that of the BitTorrent? protocol and the Peer-to-Peer networks that utilize it. Users of BitTorrent? are notorious for being selfish [Locher06], and trying to get a free-ride, that is downloading without sharing or limiting the amount they upload in order to maximize download bandwidth due to incentives built into some BitTorrent? client software to share \cite{bittyrant, incentivesbt}, cripples the service.

As users have been found to be selfish \cite{freeriding,msoftbt} when using BitTorrent? and the benefit of BitTorrent? is only realized when there are large amounts of seeders, download speeds will not reach the user's max available bandwidth. If a user needs to wait several days to weeks to download a file, chances are the download will not end up being completed because the relatively few seeders (depending on popularity of the content) have disconnected \cite{modelingbt} and the transfer ends up being terminated or paused indefinitely. Content is typically seeded when it is new \cite{modelingbt} and as time goes on the number of seeders decreases as do the number of leechers or downloaders because less seeders means slower download times, depending on the available bandwidth of the seeders. Additionally, \cite{modelingbt} discovered that the initial seeders of content tended to disappear when a sufficient number of other seeders began to propagate the content. However, if the file did not propagate to a large enough group of secondary seeders and the initial seeders left the swarm then the torrent died since the leechers downloaded what they needed, thus resulting in few if any peers serving up the content. Essentially, with larger files, seeders will need to stay connected longer to the swarm to ensure enough propagation. Staying connected longer to a swarm in order to seed and/or download leaves a system exposed on the network longer for ISPs to track down what is occurring or for covert users hired by copyright holders to build a stronger case for copyrighted content being shared illegally.

As seeders need to stay connected longer to a swarm in order to distribute larger sized content, ISPs will begin to feel a strain. ISPs typically oversell an area they service with the assumption that everyone in that area will not be using the maximum allocated bandwidth to the pipe. However, as users begin to maintain peak bandwidth levels in order to seed and or download high-definition content, ISPs will begin to feel the drain on their systems. Disregarding net neutrality for a moment, these heavy bandwidth users become detrimental to performance and disrupt service for other customers. We argue that this essentially shifts the arms race between copyright holders and cracking groups over to ISPs and their users. ISPs, such as Comcast, have been caught shutting off or disabling services for users when a mystery value for their bandwidth cap each month is reached\cite{bwcap}. ISPs are essentially in a better position to stop the proliferation of content with a large memory footprint since it ends up putting stress on their system and affects the service of other paying users.

Legitimate Content Distribution

Shifting from illegal file sharing to legitimate digital distribution, take for example iTunes and its video download service. The videos being downloaded are sub-par when comparing the quality and size of the content to that of the physical medium. Compression is used that attempts to balance size and quality such as the MPEG family of compression, xvid, divx, etc. The sizes are significantly less than that of a rip from a high definition disc and the quality is not nearly as crisp as that of a high-definition physical medium \cite{hdcomparegiz} because it is costly in terms of time to serve high definition content up to a large user base. As cited by \cite{bitrates} and \cite{hdlie} the average bit-rates for popular media are found in Table \ref{table-brates}. While different codecs are able to compress video down to different sizes that allows for a more natural looking HD picture, it is still not an HD signal in the true sense, since frame size is not the sole determining factor of high-definition \cite{hdlie}. Attempting to distribute a high definition movie with a bit-rate of 40Mbps to several hundred or thousand users in a timely manner is not feasible with the current networking infrastructure.

\begin{table} 
\renewcommand{\arraystretch}{1.3} 
\caption{BIt-Rates for Popular Media} 
% percentage of connections destined for port
\label{table-brates} 
\centering 
\begin{tabular}{||c|c||} 
\hline 
\bfseries Format & \bfseries Bit-Rate\\
\hline\hline 
AppleTV HD & 4 Mbps \\
DVD & 5-8 Mbps \\
XBOX360 & 6-8 Mbps\\
Blu-Ray & 40 Mbps \\
\hline 
\end{tabular} 
\end{table} 

Currently, broadband in the United States as defined by the FCC according to their website is 200Kbps or 0.2Mbps in at least one direction, although recently it has been reported as having been raised to 768Kbps or 0.768Mbps \cite{fccupgrade}. A typical user finds that their upload bandwidth is significantly less than the minimum FCC requirement since the requirement for "broadband" is only for a single direction. The user is more than likely to find that their broadband download speed will be a minimum of 768Kbps and go up from there depending on the level of service they pay for and the ISPs in their respective regions.

Disadvantages of a natural DRM

The disadvantages of using a natural based DRM result when bandwidth capacity catches up to content size. Recently, a major cable ISP, Comcast, decided [[http://www.startribune.com/business/17240049.html[to begin rolling out a 50Mb (down) /5Mb (up) stream service to customers in the Twin Cities area]]. A typical Blu-Ray rip of 30GB will now take only only 80 minutes to download assuming that the speed of the seeders of the content can provide a download rate of 50Mb/s to a user. The time it would take a user, with a 5 Mb upstream, to seed the same 30GB video file would be just under 14 hours, once again assuming no overhead and/or bottlenecks along the connection. However, this speed comes at a huge cost with the new high-speed Internet Service costing a consumer $150 per month, nearly double the price of current plans available, "which may not be as fast as advertised because of unpredictable data bottleneck on the Internet."

Future

Internet speeds for the general consumer will increase. However, there is no guarantee that the symmetry will even out, the US has a history of asymmetric bandwidth \cite{apbwasym} and barring a complete overhaul of the entire infrastructure it could remain like that for some time. Yet, as bandwidth increases it is a good assumption to make that content size will increase as well. Currently, an uncompressed 60fps 1080p movie clocks in at 3Gbps, which is a far cry from the 40 Mbps that Blu-ray offers.

The ceiling is still high for the file-size of content available to consumers, and there will also be increases in frame size such as UHDV requiring an even higher bit-rate to maintain quality for ultra high-definition. As all of these factors continue to increase, the idea of using a natural DRM can still be supported.

There are really only two roads to go down as far as DRM is concerned. The first road involves continuing the arms on DRM with copyright holders and a small-sub of users who crack the DRM. The second road, involves removing artificial DRM and using the natural file size of high-definition content as a form of DRM making it economically infeasible for typical end-users to share since the cost of higher speed internet will continue to increase, yet the asymmetry will continue to exist, at least in the United States until enough bandwidth provisioning can be done, which may require a complete overhaul, Clean Slate design approach.

DVD, still the most popular format, has had its DRM broken since the beginning \cite{dvddecss}. Most consumers either know somebody that can rip a DVD or know how to do it themselves. Essentially DVD's DRM has become a formality, and the current state of high-definition DRM is approaching that stage as well.

References

• [Gutmann07], P. Gutmann, "A Cost Analysis of Windows Vista Content Protection", Jun 2007.
• [Locher06]

Paper History

This paper was originally submitted to the New Security Paradigms Workshop for 2008. While the paper was not accepted for inclusion in the workshop, we thought the discussion to be gleaned from publicly posting the paper would be worthwhile. The reviews from the workshop can be found here along with a brief response to the reviewer comments.