Video QoS in triple-play networks
By Christer Bohm, Net Insight AB -- A comparison of SONET/SDH-based NGN transport and packet-only approaches may surprise you.
Net Insight AB
The integration of live television and associated video services into a telecom network poses a number of challenges for service providers. Perhaps the most important requirement is 100% quality of service (QoS), as consumers now expect this standard from MSOs and satellite providers. Video (especially video on demand), multi-channel broadcast TV, and HDTV, place greater demands on a network than voice and data. So far, the existing ATM/SONET networks and the currently deployed next-generation IP-based networks have not met the QoS challenges once the networks are required to scale to mass deployment.
As the telcos have declared that successfully delivering triple play (video, voice, and data) is a competitive necessity, this article will consider the leading technology alternatives that are being implemented by today's service providers for triple-play networks: packet only and SONET/SDH-based next-generation network (NGN) approaches.
Significantly higher requirements for video
Video has significantly different QoS requirements than data. Even the most demanding data application can cope with jitter and some degree of packet loss. However, video over IP has very strict requirement for low packet loss, in the range of 10-9, which in practice means that packets can only be dropped as a result of bit errors and not congestion. This low packet-loss rate must also be achieved with low delay, which makes excessive buffering not a viable solution (Figure 1). These requirements will be more challenging as the number of media services increases, starting with standard-definition television and followed by more advanced services such as HDTV, video on demand, etc.
An IP-based network might meet the required QoS requirements in small deployments. Congestion, aggregated jitter, and delay depend on the number of packet hops the traffic passes in the network, the load on the links, and the number of traffic sources (Figure 2). In a commercial deployment, maintaining a level of QoS consistently acceptable to the paying customers becomes a nightmare. This makes scalability a major parameter in any successful triple-play service rollout.
Early trial and deployments of triple-play services initially relied on Layer 2 approaches. When these methods were found inadequate, telcos then moved to Layer 3. Still, service providers implementing IP-based triple-play equipment have been plagued with problems due to the low quality of video transport, scalability issues, high packet loss, and cumbersome network operation. These problems are just now becoming visible, since the early implantations have only been required to deliver a small amount of video to a limited number of subscribers.
In some installations, these problems have been addressed in the short term by separating the video traffic from the bursty data traffic and significantly limiting the number of hops. Also, networks with heavy overprovisioning have been deployed. These solutions tend to be very expensive and only scale to small to medium sized deployments. Also, they are extremely difficult to trouble shoot.
Optical convergence using SONET/SDH-based NGN
The transport characteristics required for video are very similar to the ones SONET and SDH offer. These networks have a poor reputation based on their limits for datacom transport, such as very static capacity allocations, cumbersome provisioning, problems with stranded capacity, and expensive interfacing with datacom equipment.
To the relief for operators planning to introduce video services, many of these problems have been solved with SONET/SDH-based NGN transport equipment, making this technology an excellent way of curing the network problems for triple-play services. As a synchronous technology, SONET/SDH preserves timing, which results in no packet loss regardless of the number of hops. Using an enhanced SONET/SDH platform, each traffic stream on the network is allocated the bandwidth it needs, resulting in a cost-effective network with high bandwidth utilization (Figure 3).
The third way
Many telcos have already announced their intention to migrate their networks to a complete IP infrastructure. This long-term strategy makes sense for networks that carry both data and voice services. Therefore, even with the known video limitations, operators would greet an approach that ignores IP-based transport with some skepticism.
Perhaps a better approach to video transport would be to incorporate the best characteristics of IP and SONET/SDH-based NGN transport into a single architecture. For example, a combination of SONET/SDH-based NGN transport with IP routers handling the aggregation and distribution of traffic with guaranteed transport would offer service providers the best of both worlds. The number of IP hops would be limited by using SONET/SDH-based NGN transport to transport the video to the router or the DSLAM.
Additionally, as further proof of this approach, media and TV companies use this architecture for global transport of uncompressed video as part of their mission-critical production and distribution to and from affiliates and remote locations. The QoS and bandwidth requirements for broadcast video greatly exceed those for home triple play.
True best of breed
"Best of breed" has become a very popular buzzword in telecom. The idea of combining multiple network technologies into a single solution often ends up of being more of a goal than a commercial reality. Today, commercial triple-play installations in the United States are benefiting from a combined SONET/SDH-based NGN and IP router architecture. These networks have been delivering trouble-free, high-quality triple-play services with more than 150 IPTV channels, high-definition, and video on demand for over two years.
As triple play networks begin to become ubiquitous, this approach will be implemented on a large-scale basis. To unseat incumbent TV providers, consumers will demand a solution with guaranteed 100% QoS as non-negotiable starting point.
Christer Bohm is CTO of Net Insight AB (Stockholm, Sweden).