Video layer quality of service
Controlling and managing video quality as HD options expand.
Spurred by DirecTV’s 2007 declaration that it will be the world’s first video service provider to reach 100 HD channels, cable operators are moving rapidly to create additional bandwidth, not only to carry dozens more linear HD channels, but also to provide hundreds, and eventually thousands, of HD VOD titles. The video quality bar is simultaneously rising due to the mass consumer adoption of large HDTV displays and the growing popularity of Blu-ray.
Leveraging a concept from IP networking, Video Layer Quality of Service (QoS) involves creating minimum and maximum video quality values at the service level. One technique for achieving that goal is using variable bit rate (VBR) video coding.
VBR has several advantages to recommend it, first among them that the technique is a cost-effective means of increasing bandwidth efficiency without sacrificing video quality. Similarly, Video Layer QoS allows the optimization of bandwidth efficiency while guaranteeing the QoS in a sustainable manner throughout the various switching, multiplexing and splicing stages of video communications networks.
The North American market for video subscribers is becoming increasingly competitive and fragmented, with cable, DBS, telco and Internet video providers all jockeying to gain a bigger piece of a growing pie, and HD is one of the key battlefronts. Since most major service providers are moving toward virtually unlimited libraries, the next step in competition is to provide better video quality.
We define Video Layer QoS as the establishment of video quality levels at the content origination or delivery site, combined with the process of sustaining these levels all the way to the consumer viewing environment in the most bandwidth-efficient manner possible.
A Video Layer QoS solution provides:
1. Excellent MPEG-2 or MPEG-4 video quality, at greater bandwidth efficiency than today’s encoders, from content origination all the way to the set-top box.
2. Consistency (equalization) of a service provider’s video quality across the digital broadcast, SDV, VOD and network PVR (e.g., Start Over) categories.
3. The ability to assign different quality levels to different classes of assets (e.g., HD VOD, PPV), or even individual assets (e.g., the Super Bowl), at the discretion and under the control of the content provider or operator.
4. Sustenance of the pre-calibrated video quality levels in a cost-effective, non-disruptive and backward-compatible manner throughout the various multiplexing stages, including local and addressable ad insertion.
5. Ongoing measurement and reporting capabilities, as well as the ability to perform auto-healing of individual digital video streams.
Video compression systems remove unnecessary bits from digital video signals, ideally with minimal impact to the consumer viewing experience. The ubiquitous audiovisual coding standards of MPEG-2 and MPEG-4 AVC (H.264) are designed specifically for this purpose. Traditional encoders, however, contain relatively simplistic representations of the human visual system, while headend rate-shaping devices implement quantization techniques based primarily on squeezing signals into a target bandwidth.
In Figure 1, the images show the same picture compressed with three different methodologies. In the first image, a “compression by quality” technique is used, in which a sophisticated objective video quality measurement system, representing the Human Visual System, is leveraged to minimize compression artifacts relative to the source. In the second image, a typical MPEG-2 encoder is used. In the third image, pre-processing and high-frequency pixel filtering techniques are used, in addition to the traditional compression methods.
The first image appears noticeably sharper and cleaner than the other two, showing neither the blocking artifacts of the second image, nor the blurriness or loss in detail of the third image. All other things being equal, such an improvement in perceived quality can be made possible if the video quality has been exhaustively analyzed and measured as part of the video processing and multiplexing solution.
Figure 2: Video processing.
PROCESSING BY QUALITY
A method of video processing, involving “closing the loop” with an objective video quality measurement system, can allow signal quality and bit rate efficiency to be significantly improved, while providing Video Layer QoS through the re-processing, re-multiplexing, VBR-to-CBR conversion (or vice versa) and splicing stages.
Step 1: Select or devise a video quality measurement technique
An ideal technique involves: 1) high correlation with subjective video results; and 2) optimization for the MPEG-2 and MPEG-4 AVC standards. Since pre-existing methods (e.g., PSNR, VQM, SSIM, etc.) do not meet these criteria, Imagine created a new technique called ICE-Q(r).
Step 2: Video processing
The video processor must “close the loop” with the selected objective video quality measurement method. This involves processing every frame and every macroblock as part of the selection of a constant video quality “level” or “grade.” By definition, this signal is VBR, since the activity and complexity vary over time.
In great contrast to today’s encoding or rate-shaping methods, the video processor is configurable to a pre-calibrated quality level rather than a maximum, minimum or average bit rate. A recommended method to guide this process is to use a mathematical scale, such as 1 to 100, rather than more subjective groupings, such as “good,” “bad” or “average.”
Step 3: Calibration
All of the available signals or video assets are processed using the video processor from Step 2, employed at various selected quality grades. One can define a minimum of two quality grades in a similar fashion to the QoS utilized in IP networks as follows:
1. QGa – target average video quality grade; for example, “96.”
2. QGb – Guaranteed or minimum allowed video quality grade; for example, “90.”
QGa can be defined as “no material degradation” (NMD), which means even expert viewers can rarely discern differences from the source. QGb can be defined as the quality level at which, the vast majority of the time, ordinary viewers can’t discern differences from the source.
A good practice for delivering the signals is to use QGb frames or packets less than 1 percent of the time.
Step 4: Statistics
Process all of the target channels or video assets at QGa and QGb and gather statistics for at least 24 hours, or preferably for one week. Measure the respective bit rates per second and create two vectors, one for each quality grade.
Step 5: Lineup
Determine the optimal digital service combination per multiplex with a goal of providing QGa quality, on average, and never less than QGb. Relevant equations can then be used to optimize the multiplex lineup using the bit rate measurement statistics.
Step 6: Statistical multiplexing
Process the signals in real time at multiple quality levels, and design the statmux device to always select the highest-quality grade possible under the maximum channel bit rate constraint. The proportion of null packets should be as close as possible to 0 percent.
Figure 3: Statistical multiplexing.
THE BOTTOM LINE RESULT
Video Layer QoS provides an unprecedented level of control for a system operator or content provider, all the way from content origination to the set-top box. Assuming the IP and MPEG-2 transport layers are intact, this capability opens up new possibilities for ensuring video quality not available with previous digital or analog delivery solutions. Technically, Video Layer QoS means maintaining the pre-determined quality requirements (QGa and QGb) through the communications delivery network, including sustainability through the various re-multiplexing, splicing, encryption, edge statistical multiplexing and VBR-to-CBR conversion for services such as Start Over and SDV.
There are two main approaches for ensuring video quality of advertisements during ad insertion. The first approach involves pursuing the highest-quality possible for the ad, even at the expense of the underlying digital services not containing ads at the same time.
The second approach involves equalizing the ad quality to the underlying stream quality to the extent possible. In this case, the ad is multiplexed at QGa, or at the eventual average quality grade of the primary stream. The ads should be processed and stored on the ad server at the maximum possible bit rate and quality level, providing downstream flexibility.
CBR for VOD and SDV
Applying Video Layer QoS to the conversion of VBR signals to CBR (for SDV and nPVR applications) is relatively straightforward, with the quality levels being calculated in advance at the content origination site.
Figure 4: Edge statistical multiplexer.
A state-of-the-art edge statistical multiplexer can increase, by up to 50 percent, the number of streams per QAM channel without quality degradation for VOD, nPVR, SDV and IPTV applications.
In order to simultaneously maintain the Video Layer QoS and optimize the bandwidth efficiency, it is important to also involve the Edge or Session Resource Managers (ERM/SRM).
This involves allocating the bandwidth according to each service’s effective bit rate (BQ), and then load balancing the quality across the switched QAM channels, thereby guaranteeing Video Layer QoS. This method ensures the best quality at any given bit rate for edge and switched applications, including VOD, nDVR, SDV, nPVR, switched unicast and addressable ad insertion. The video quality will be significantly higher than today’s capped quality at 3.75 Mbps. Also, bandwidth efficiency can be increased by up to 50 percent, allowing 15 SD or 3 HD VBR streams per QAM channel.
Using a similar approach, combined with standard IP QoS mechanisms, constant video quality and Quality of Experience (QoE) can be achieved for video services delivered over the Internet and to mobile devices.
The cable industry is in the midst of a dramatic transformation toward an increasingly competitive and complex environment. Multiple categories of digital TV services will co-exist on a unified platform, including digital broadcast, VOD, SDV, nPVR and Internet video, each of which will encompass SD and HD signals.
This evolving, comprehensive suite of services and architectures must be presented in a transparent and convenient manner to consumers, who now have multiple choices for their service provider. In this new environment, a key consideration and a competitive differentiator is the ability to provide true Video Layer QoS, combining control and optimal video quality across all categories with the utmost in bandwidth efficiency.