June 2005

Fabrice Beer-Gabel, BigBand Networks
Reprinted with permission of the United States Telecommunications Association

1.0       Introduction

Telecom carriers facing more competition from cable, wireless and satellite operators are turning to video services while striving to leverage the investments already made in their copper, fiber or coaxial cable plants. The opportunity to build a video operation from the ground up presents telcos with a unique opportunity to make the most economically and technologically sound decisions without having to support legacy video infrastructure. 

Any choice that a telco makes should have the goal of providing a compelling user experience while minimizing capital and operating expenses. Additionally, any network deployment must be versatile enough to support a broad array of new services, provide seamless connectivity, and scale over time.

This chapter outlines the role of video-aware networking to meet these objectives and outlines key considerations and alternatives for the design of a robust video network that supports a compelling user experience and optimizes for profitability.

2.0       Video-Aware Networking

Video networking is distinctively different from voice and data networking, and a backbone architecture designed to provide legacy services over IP to subscribers may be inadequate for delivering competitive video services. Video has unique characteristics such as a requirement for real-time delivery, the need for low inter-packet jitter, and greater consumption of bandwidth than is typical with other media. Optimal performance requires media intelligence, meaning that network elements transporting video across the network must have the capability to inspect and process content in order to make the best routing decisions.

Not only should video awareness extend to knowing where each video packet needs to be routed to, and in what order they need to arrive, but also to knowing how individual programs should be treated so as to maximize picture quality, minimize bandwidth requirements, and provide superior service reliability. Achieving these goals enables a carrier to increase competitiveness and enhance subscriber satisfaction.  A video-aware network also serves as the foundation on which new services can be added to meet future technical and commercial requirements.

 

Figure 1: Video-Aware Networking Map
(click to enlarge)

Figure 1 outlines some of the key capabilities of a video-aware network, each of which is discussed in this chapter.

Video-aware networking requires the combination of video processing and networking functions in a carrier-class programmable platform. A video networking platform leverages the installed infrastructure to provide a differentiated user experience, support scalability and competitive service reliability.

3.0       Key Video-Aware Functionality

Some of the functionality described briefly in the previous section warrants deeper discussion. These include the role of service-aware switching, digital advertising, bandwidth management, FEC, de-jittering and encryption. First, however, the major benefits of an integrated video-aware networking platform are discussed.

3.1       Benefits of Platform Integration

The video-aware networking platform integrates rich media processing with robust transport to provide high-quality video services at lifeline reliability. A programmable platform takes advantage of technologies such as FPGAs and DSPs that provide both high performance and maximum flexibility for network evolution using new services such as HDTV and network-based PVRs, new compression standards, including MPEG-4 H.264 and Windows Media VC-1, and other software upgrades. Programmability includes the ability to remotely load software upgrades.

Carrier-class service reliability is ensured using a robust chassis with built-in redundancy. Additionally, a modular design further minimizes video channel interruptions by reducing the number of separate network pizza boxes or video processing appliances that each introduce potential failure points and cabling problems.

Seamless, end-to-end management communications using open protocols such as SNMP, enable a program to be monitored from a single location, and provide a global view of network performance.  A video-aware networking platform also allows programming changes to be made using a simple drag-and-drop operation in the management GUI, avoiding the need for time-consuming, and error-prone re-cabling. 

3.2       Service-Aware Switching

A key advantage of a video-aware architecture is continual knowledge of the state of each video program transported across the backbone, leading to the ability to make switching decisions accordingly.  Figure 2 shows the necessary capabilities of a video networking backbone. 

The first capability is program-level redundancy.  In a redundant topology, two headends acquire identical feeds from sources such as satellite, local antennas and off-air sources, and transport the content downstream to a regional headend. Within the regional headend, a video networking platform constantly examines the quality of each program and makes an automatic protection switch to the secondary source if a failure or significant degradation is detected in any of the programs on the primary feed. Program-level redundancy means that degradations that are only detectable at the media layer, such as packet identifier (PID) drops and other changes in the video payload, can be corrected automatically and without impact on the subscriber viewing experience. 

A second capability is real-time slate insertion, which happens automatically in a video-aware backbone should a blackout, or other service interruption, occur. A typical slate informs subscribers that a service outage has occurred and is being addressed. Slate insertion reduces subscriber calls to service centers, lowering the costs incurred in responding to customer complaints.

 

Figure 2: Key Service-Aware Switching Functionality
(click to enlarge)

A third capability is support for emergency alert system (EAS) broadcasts that allow emergency authorities access to various communications channels for the public good in times of crisis. Receivers are commonly placed at a regional headend, and have the capability to signal the headend to start broadcasting the EAS messages at the appropriate time. Once the headend begins broadcasting, the video-aware platform automatically switches the channel that a subscriber is watching to the EAS message.

3.3       Digital Advertising

Not only does advertising represent a significant revenue opportunity for a telecom carrier deploying video services, it helps mitigate business risks by diversifying revenue sources with a valuable and differentiating community service to local businesses. An operator planning the deployment of advertising services should consider the multiple implementation options available, and understand their respective impacts on profitability.

Although the first generation systems for splicing advertisements into programs streams were analog, telcos can greatly benefit from implementing an all-digital insertion architecture to reduce implementation costs. The massive migration to digital services has brought with it digital commercial insertion standards and technologies that introduce efficiency and flexibility in the ad-insertion process. In a digital insertion system, a television channel is received from a programmer as a compressed digital stream, currently based on MPEG standards. Within the stream, specially placed cue markers provide the exact timing location of advertising opportunities or "avails" for the program. The cue messages are specified by the Society of Cable and Telecommunications Engineers (SCTE) 35 2001 standard and multiplexed into the programs sourced at the headend.

Figure 3: Digital Advertising Architecture
(click to enlarge)

The cue tones are detected by a splicing device that in turn engages in the negotiation of one or more commercial insertions. This negotiation follows a standardized protocol, SCTE 30 2001, enabling the ad server to stream commercials to the splicer which, in turn, inserts them into the program streams. The media processing capabilities present in the splicer allows this insertion to occur in both a precisely timed and visually seamless fashion. The splicing process is handled entirely in the compressed digital domain, eliminating the expense of decompressing and re-compressing the program.
Current field deployments are based on the MPEG-2 algorithm but the SCTE standards also enable implementation of digital ad splicing on advanced compression algorithms such as H.264.

As digital television migrates to on-demand personalized services, the TV advertising models will follow the same trend.  Subscriber communities can be segmented into ad zones according to geographic or other demographic distribution to support targeted advertising.  The combination of switching and media-awareness in the network enables a programmer to insert different commercials into program streams addressed to the different zones.  Zoning of broadcast services will become increasingly granular, eventually enabling addressable advertising to target individual subscribers.

3.4       Bandwidth Management

The ability to add subscribers to a network along with offering a wide breadth of services are a central requirement of a competitive telco video strategy.  Video is a bandwidth-hungry media and several techniques can be considered for mitigating bandwidth constraints and limits on the number of incremental subscribers that can be served economically. These techniques apply across the access networks and transport schemes outlined in section 5. They include rate shaping of statistical multiplexes consisting of multiple program transport streams in both standard and high definition, and advanced compression approaches such as H.264. The appropriate use of transcoding, in which video in one compression format is converted to a different video format, can significantly lower bandwidth requirements. This includes encoding of video sources in bandwidth-inefficient analog formats to digital.

Each of these techniques must be implemented with sophisticated algorithms that maintain the highest video quality in order to ensure that the user experience is maximized. These techniques also include the ability to prioritize media processing on a per service basis. 

Other bandwidth management functions include the use of switching and multicasting to transmit only viewed programming on the network.  Switched video service, of which switched broadcast is an example, provides a means to offer subscribers a wide range of local and niche program with little or no burden on network capacity. SAP (second audio program) splitting, a function that conserves bandwidth by sending only the accessed audio program to a subscriber, instead of sending all audio programs that accompany a video stream, minimizes bandwidth requirements on the access network. 

3.5       Forward Error Correction

Packets lost or corrupted during video delivery noticeably affect a subscriber’s viewing experience because the resulting artifacts can be very visible, and may be sustained through multiple frames. It is essential therefore to design a network that provides maximum packet error resiliency since the real-time nature of video prevents packets being re-transmitted. Forward error correction (FEC) is a well-known method for improving error resiliency.

Different types of FEC algorithms have been developed, including Reed Solomon, Hamming or XOR, with interleaving options. The use of Reed-Solomon is sufficient in networks that use quadrature amplitude modulation (QAM) for last-mile content distribution, such as FTTP networks with an RF overlay, or an hybrid fiber coax (HFC) network.  IPTV presents a challenge, however, because packets can be discarded due to transmission bit errors coupled with UDP checksum errors at network elements on the backbone. The challenge is increased by analog interference errors such as analog ring tones, on the access network. The particular structure of H.264 or VC-1 algorithms makes the problem even more sensitive, since these algorithms rely on fewer reference frames than MPEG-2 in order to maximize bandwidth efficiency, potentially resulting in longer recovery times.

The optimum implementation of FEC minimizes utilization of network resources, while maximizing the quality of the subscriber experience. Table 1 summarizes some of the key considerations when implementing FEC in a video network.

Objective	Motivation
Maximize error resiliency	Recovery from specific errors that occur due to network architecture
Minimize bandwidth overhead	Leave maximum bandwidth for video delivery in bandwidth-limited environments such as ADSL2+ FTTN architectures
Minimize latency	Maintain system user-friendliness in situations of channel changing or control of an on-demand session

Table1: FEC Implementation Objectives

Video-awareness in the network provides an implementation of these solutions that can be complemented by an in-depth understanding of the video payload structure to further optimize the error resiliency equation.

3.6       De-jittering

Video compressed using MPEG algorithms is sensitive to jitter because the transport stream carries the timing information used by a set-top box (or other type of receiver) to reproduce the baseband program, and variation in the arrival of packets can cause loss of synchronization. In an end-to-end digital television system, the clock is typically locked to the incoming baseband video stream and drives a counter that generates a program clock reference (PCR).  Minimizing PCR jitter is important in maintaining a robust and high-quality end-to-end signal. ISO standard 13818-1 specifies a maximum PCR jitter of 500 ns, but this value specifically does not include jitter that can be caused by UDP/IP encapsulation and network transport by IP or other protocol.

Without an effective de-jittering mechanism, excessive PCR jitter can cause packet deliveries that violate the buffer models specified by MPEG, and more importantly can prevent the decoder from accurately regenerating the clock signal required to reconstruct the baseband signal. Televisions that encounter this type of impairment will generally be unable to lock to the corrupted “color burst” waveform that precedes the delivery of a field of NTSC video, typically resulting in TV picture becoming “black-and-white. In instances of severe jittering, loss of service may result.

A video-aware networking platform has the ability to cope with high network jitter by intelligently correcting the PCR value in the transport packet and/or shaping the flow of the packet traffic as it transits through the device.

3.7       Encryption

Telecom operators can restrict access to content using a conditional access (CA) system. There are various reasons for restricting access to content, including the need to enforce payment by the end user for viewed services, the need to restrict access to programming in a particular geographical area because of program rights considerations, or the need to facilitate parental control. With the emergence of interactive TV and home content storage, digital rights management (DRM) capabilities are added to further protect content copyrights and enforce new on-demand service models.

A variety of standards-based and proprietary CA systems are available. The Digital Video Broadcast (DVB) consortium provides suggestions for a set of conditional access techniques on different types of networks, including IP-based networks.  A CA system has three major functions: (1) scrambling and descrambling of content; (2) encryption and decryption of the control words used to scramble and descramble content; and (3) creation of entitlement control messages (ECMs) and entitlement management messages (EMMs) that are used to communicate information between the scramble and descrambler. These messages can also be used to access permissions and provide other information to a subscriber’s set-top box.

A video-aware networking platform has the capability of implementing various scrambling algorithms and interface with various CA systems over open standard interfaces.

 4.0       Subscriber Quality of Experience

The overall performance of a video network from the subscribers’ perspective, also referred to as the subscriber’s quality of experience (QoE), must be considered when designing the backbone. Picture and audio quality are perhaps the two most distinguishable metrics that affect a subscriber viewing experience, but attention should also be paid to channel switching time, the frequency of service interruptions, the breadth and availability of personalized content that a subscriber can view, and other metrics. QoE targets should be established for each video service and be included in the network design.

A video-aware networking platform offers a telco the ability to substantially improve the subscriber QoE by providing mechanisms for enhancing service reliability, minimizing channel switching time, and offering a wide range of local and niche programming content. Scalability and efficient use of bandwidth enable new services to be provided to a subscriber quickly and cost-effectively.

5.0       Network Architecture Options

A telecom carrier designing a video network architecture faces several choices and must make decisions based on specific technical and business opportunities and constraints.  In some cases the best business model results from deployment of a hierarchy consisting of three tiers that span content acquisition at one or more headends, content localization and protection at regional headends, transport over a backbone to the network edge, and delivery to subscribers over an access network. In other instances, however, a telco may decide to focus on only two tiers, backbone transport and last mile distribution. The decisions made are typically based on subscriber coverage and redundancy considerations.

This chapter concludes with an overview of the important video-aware networking functionality required at each of type of office.

5.1       Video Networking Overview

A video backbone logically begins at the content acquisition point, commonly referred to as the headend. A variety of sources of content are typically found within the headend, such as satellite dishes receiving content that has been broadcast over hundreds of miles, antennas picking up local content and local landline feeds. Servers storing content such as advertisements and on-demand programming are also found at the headend. It makes sense for networks with large and widespread subscriber communities to separate content acquisition and content localization functions via the creation of super headends (SHEs) and regional headends (RHEs).

Redundancy is paramount to the SHE design given its subscriber coverage. For maximum reliability, operators can interconnect two geographically dispersed SHEs with program level redundancy capability, as discussed in section 3.2.

Figure 4: Generic Video Network
(click to enlarge)

Content acquired at SHE is transported over a video backbone to a video hub office (VHO). Gigabit Ethernet delivery of UDP/IP packets is the most common method for transporting programming from a SHE, and DWDM or CWDM ensure maximum utilization of the optical fiber plant.  

The key roles of the RHE are content localization and protection. A video networking platform at the RHE splices commercials into live content. Insertion of local content such as broadcast channels and PEG programming to video streams that have been transported across the backbone via different multiplexing techniques, also occurs at the RHE.

The function of the central office, or video serving office (VSO), is digital distribution of content to remote terminals from which homes are served via various types of physical access network such as copper, optical fiber or, in some cases, coaxial cable. A carrier delivering video over copper will face bandwidth limitations not encountered with fiber plant, but upcoming bit rate adaptation techniques applied to video content promise to help mitigate these issues.

As subscriber penetration increases and services become more personalized, some of the content localization functions currently performed at the RHE will migrate to the central office.

5.2       Access Network Considerations

Three different types of last-mile network are available to telcos developing a video deployment strategy. Many of the major ILECs have provided, or are planning to provide, IPTV over copper-based access networks.  Some ILECs are building FTTP networks and using RF/QAM to deliver video content into subscribers’ homes. Other telcos are using coaxial cable in the access network, or all three types.

5.2.1   Copper Access Networks

The most common method for providing last-mile connectivity to an individual subscriber’s residence is by copper twisted pair, using ADSL (and more recently ADSL2+). The inherent bandwidth limitation of copper puts pressure on the performance of compression formats such as MPEG-2, and advanced video compression protocols schemes such as MPEG-4 H.264 will particularly benefit this type of network. Telcos can leverage their network infrastructure to provide an IP-based switched service where bandwidth limitations are driven by the number of simultaneous viewing devices in the home, rather than by the number of channel on offer. In an IPTV scenario, each IP set-top box in a subscriber’s home receives an individual single program transport stream (SPTS).

5.2.2   Fiber Access Networks

Some carriers are deploying, or have already deployed, optical fiber to the home, which offers significantly higher bandwidth capacity and flexibility than copper, but at a much higher install cost. Architectures that use passive optical networking such as BPON and GPON are proving viable.

Options for delivering video over fiber include an RF overlay, an IP/RF hybrid, or a  pure IP implementation. Quadrature amplitude modulation of video streams over RF is typically the preferred method for delivering video to CPE in an FTTH scenario. However, some  fiber deployments leverage WDM to provide a subscriber with QAM/RF video on a 1550nm wavelength, and IP-based services on a 1310nm wavelength. A pure IP over fiber implementation can also deliver robust and cost-competitive video service by sharing the available bandwidth among all services.

An RF overlay scenario leverages proven video technologies and, furthermore, enables the deployment of an analog tier that allows the use of less costly set-top boxes. Similar to IPTV, an RF-based access network can benefit from a   switched video model that uses innovative video-aware techniques for increasing subscribing viewing choices while reducing bandwidth consumption.

With an RF overlay a set-top boxe in a subscriber’s home receives a multiple program transport stream (MPTS) carrying a subset of the video services on offer. The service selection is implemented at the CPE instead of in the network, as is the case in an IPTV scenario.

5.2.3   Coax Access Networks

In some instances a telco has deployed or inherited coaxial cable connectivity to subscribers, and uses QAM/RF to deliver content to subscribers. Although coax networks were originally designed for broadcast delivery, they can be augmented to fully support switched services such as switched broadcast and video on demand. Coax networks also support IP-based video distribution via the use of DOCSIS (data over cable service interface specification) signaling. The forthcoming DOCSIS 3.0 definitions with next generation CMTS (cable modem termination system) platforms should be well suited to IPTV. 

6.0       Conclusion
Video networking is distinctively different from voice and data networking, and a network architecture designed to support voice or data will probably be inadequate for delivering video. This is due largely to the fact that video has unique characteristics that include the need for real-time delivery, a requirement for low inter-packet jitter, and greater bandwidth consumption than other media. In order to provide competitive, compelling video service, a carrier should build video-awareness into the network backbone. Doing so, requires video-aware networking platforms smart enough to examine content, and process and route traffic for best results. Subscriber quality of experience provides the metrics to determine the success of a video network design. 
A video network built using media-aware platforms will consist of key functionality such as service-aware switching, support for digital advertising, provisions for FEC, and de-jittering and transcoding capabilities. The ability to support carrier-class network availability, subscriber addressability, scalability and content protection using encryption are essential.

For more information about BigBand Networks, please contact:
Corporate Headquarters
475 Broadway, Redwood City
CA 94063, USA
(t) +1 650 995 5000
(f) +1 650 995 0060

European Headquarters
Abbey House, 18-24 Stoke Road
Slough, SL2 5AG, UK
(t) +44 1753 722 146
(f) +44 1753 722 145

Asian Headquarters
Suite 1422, Prince’s Building
10 Chater Road, Central, Hong Kong
People’s Republic of China
(t) +852 3151 7304
(f) +852 3151 7385

Or go to our website: www.bigbandnet.com