Video-capable DSL: Why legacy DSL platforms are becoming obsolete

Dec 8, 2004 12:00 PM, By Geoff Burke

Eight years have passed since the RBOC Joint Procurement Consortium made its landmark investment in ATM DMT-based DSL equipment. Since the mid-‘90s, carriers have been deploying DSL in their access networks to deliver Internet access and enable applications such as e-mail, file sharing, Web browsing, and e-commerce. From a processor, component and backplane perspective, ATM-based DSL devices were cost-optimized for these relatively low-bandwidth applications to allow carriers to cost-effectively build out their DSL networks and capture subscriber market share.

Today, the broadband connection to the subscriber has become a critical battleground, as wireline carriers, cable companies and other emerging broadband competitors compete to capture the next big wave of service revenues from emerging high-bandwidth broadband services--including IPTV. To accommodate these services, wireline carriers are now looking to upgrade their DSL platforms using new technologies like ADSL2+ to deliver the 15 to 20 Mb/s necessary to support multiple video streams into the home (including HDTV), multi-megabit high-speed data and voice-over-IP (VoIP) services. Wireline service providers expect that a combination of service bundles and raw demand will cause user take rates for broadband services to soar from 5% to 10% of subscribers today to 30-50 percent over the next few years. The implication is clear--wireline service providers have high expectations for these emerging services, and see the establishment of broadband relationships as key to their future growth.

Unfortunately, successful penetration could potentially lead to the undoing of service provider legacy networks. Even at low penetration rates, always-on, multi-megabit broadband services will easily overwhelm the system capacity of legacy DSL platforms. Providing video streams to three standard-definition TVs simultaneously today requires a constant bit rate of at least 11 Mb/s--7 days a week, 24 hours a day. Unlike today’s data-centric DSL applications, video is streamed to the home all hours of the day or night, whether a TV is on or not. Even traditionally manageable services like high-speed data will be challenged both by substantially increased speed requirements and new quality of service requirements driven by applications such as video gaming and IP-based competitive voice services.

Capacity is not the only factor taxing legacy DSL platforms. Increased demand for both broadband services and bandwidth requirements highlight the necessity for a simplified new approach to subscriber management and platform evolution. To rapidly capture market share with high-value services, automated provisioning will be required instead of manually provisioned ATM PVCs. To support many of tomorrow’s high-bandwidth services, a network evolution to fiber to the premises may be required--a transformation that requires wholesale change--out of most of today’s legacy equipment. Meeting the demands of the future will require new tools and support programs as well as new equipment.

Consumers and competitors have thrown down the proverbial gauntlet, and wireline service providers need to look closely at their existing DSL access architectures. The objective: to determine if their networks can support current trends, not only in terms of pure bandwidth required, but in terms of how these services affect equipment and network capacity, manageability and flexibility.

Emerging bandwidth demands

Without question, the most challenging new service to be offered over DSL access infrastructure is video. Whereas high-speed data services are “bursty” by nature, with intermittent requests for data followed by large bursts of traffic, video services are “dedicated,” requiring uninterrupted streaming for the duration of the session. Unlike data, video content changes more than 30 times each second, requiring updated content to be sent constantly. In addition, each updated packet must be delivered and reassembled at its destination in the same order in which it was sent from its source without interruption. Even the slightest delay or packet loss will result in a distorted picture--frequently referred to as “tiling” or “freezing”--that is unacceptable to most end users. Adding further complexity to this mix is that:

• Each video stream today represents a minimum of 3.8 Mb/s at today’s video encoding rates

• Typical households require at least 3 video streams to be delivered

• End users seldom “turn off” their set-top boxes, requiring these signals to be delivered 24 hours each day.

The result is more than 11 Mb/s of uninterrupted bandwidth consumed by a typical video-enabled household every hour of every day. If we were to compare this requirement to a 1 Mb/s high-speed data service in the past, over-subscribed by a typical 25:1 ratio (net 40 kb/s), we see an overall traffic increase of more than 275 times for each service subscriber. Although advanced new compression technologies like MPEG-4 and Microsoft Windows Media 9 promise to reduce the bandwidth per stream by as much as 30% to 40%, other emerging applications like HDTV and digital video recording (DVR) will in fact increase overall bandwidth requirements. Each HDTV stream today requires between 15 to 19 Mb/s of bandwidth for delivery. Even with advanced compression, these streams are likely to be in the 7 to 9 Mb/s range. Thus, the addition of just one HDTV stream into the typical household increases the overall bandwidth requirement significantly despite advanced compression.

DVR will also increase bandwidth requirements, as the total number of concurrent streams into the home will likely increase as the DVR actively records programs independent of those watched by the consumer. It is no longer a question of if HDTV and DVR will be part of the typical household services mix, but rather a question of when. These services, coupled with increasing bandwidth requirements from high-speed data services to standard multi-megabit speeds and other emerging broadband services like VoIP will continue to increase the overall dedicated bandwidth requirements to the household beyond 15 Mb/s within the next five years.

See Figure 1

Capacity issues

These bandwidth demands have compounding effects throughout the service provider network, as previously unnoticed bottlenecks either need to be removed or accommodated. As demonstrated in the discussion of emerging bandwidth demands, each video-enabled line card now carries at least 275 times the traffic that it previously needed to deliver basic high-speed data services. Even in this most basic of scenarios, the system bus between each card and the central fabric now carries 275 times the amount of traffic, resulting in multi-gigabit per second backplane requirements.

 The increasing bandwidth requirements of HDTV, DVR, multi-megabit high-speed data, and VoIP will certainly have a significant impact on network capacity, but other emerging issues like increasingly on-demand services will also affect capacity tremendously. Most wireline video deployments today utilize multicasting to minimize the overall bandwidth delivery requirements.  The rise of services like HDTV, which will introduce incremental channels into the service delivery network, and DVR, which will draw niche content from the lesser-used channels in the network, will mitigate many of the network advantages of multicasting by increasing the variety of channels regularly viewed. However, the biggest influence on network capacity will be the proliferation of on-demand service over the next decade. Each on-demand transaction counteracts the bandwidth optimization that multicasting provides by introducing new, discrete video streams into the network. Ultimately, on-demand services are likely to account for the majority of customer transactions over the next 5 to 10 years, effectively removing any bandwidth efficiencies that multicasting traditionally provided and creating an increasingly unicast network in which there is a 1:1 correlation between the number of active TVs in the network and the bandwidth per video stream necessary to serve them. Thus, whereas it may be possible to deliver 100 video channels to 20,000 video subscribers while utilizing about 380 Mb/s of network capacity (assuming each channel represents 3.8 Mb/s), delivering a video-on-demand (VOD) service to just 1% of those subscribers would require 760 Mb/s, resulting is a total video services capacity of nearly 1.2 Gb/s. It is easy to see how proliferation of on-demand services dramatically increases network capacity requirements independent of the presence of new content like HDTV. If we increase the encoding rates of on-demand content to reflect these new services as well as average take rates, we see these network capacity requirements exacerbated even further.

See Figure 2

Manageability of video rollouts

Historically, when delivering broadband services to the home consisted of merely enabling a best effort sub-megabit-per-second service to a home, plant qualification, home installation, and service provisioning required little effort--the number of vendors involved was minimal, the service requirements were low, and in many cases the subscriber was able to self-service. Unfortunately, as the sophistication of the services increases to include video, the sensitivity of the services to interference, the variety of vendors necessary to deliver these services, and consequently the integration and provisioning of these services became significantly more challenging. Because the race to provide broadband services to the home is so competitive, overcoming these challenges to provide time-to-market advantages is imperative.

Addressing these issues requires planning during the product development process, iterative product improvement and tool development that result from real-world experience, as well as the formulation of formal programs to drive out integration complexities. Whereas historical DSL platforms were never designed to deliver video services, the new breed of highly scalable multi-service DSL systems have been designed to open FSAN and ITU industry standards. Additionally, whereas traditional DSLAMs have limited diagnostic functionality because of their reliance on craft interfaces and expensive element management systems, improvements have been integrated directly into new multi-service platforms to facilitate common issues like bulk transport provisioning, rapid video deployment templates, and bulk cross-connect re-provisioning. The result is dramatically improved automation and simplicity in turning up new subscribers and managing existing subscribers.

Despite industry standards, integration of numerous vendors for video, from headend to middleware to set-top boxes, can also be challenging. Some new multi-service vendors have made significant investments not only ensuring and certifying multi-vendor interoperability, but creating extended service programs that assist in advanced services deployment, from network planning to turn-up and launch. Wireline service providers who take advantage of these programs benefit not only from dramatically reduced complexity of integration and time to market, but still benefit from all of the cost and innovation advantages of participating in an open services market.

Platform flexibility

Just as traditional DSL platforms are challenged to meet the requirements of today’s emerging applications, today’s multi-service platforms may be challenged by high-bandwidth applications of the future. Recognizing this possibility, today’s multi-service platforms have been designed to seamlessly support both an evolution to pure fiber-to-the-home infrastructures, as well as the reality of hybrid scenarios where fiber terminates in the central office, remote terminal, the curb, or even at the home. In doing so, these platforms protect the wireline carrier’s access investment by providing natural and cost-effective migration paths to enable future applications with even higher bandwidth, while still enabling services and user interface continuity across all subscribers.

Conclusion

Legacy DSL platforms were built and implemented reflecting the realities of the time of their deployment--the Internet was still in its infancy, copper was the predominant access medium, and delivery of video over the telecommunications infrastructure was almost unimaginable. Since that time, the Internet has changed the world, fiber has been widely deployed throughout the access infrastructure, and the delivery of video services by wireline providers has gone beyond feasible to become a competitive necessity. You would not expect the same vehicle that flies you 1400 miles between San Francisco and Dallas to be able to take you 275 times farther to the moon, yet this is equivalent to the capacity increases required of legacy DSL platforms to support even basic video services. In contrast, today’s multi-service platforms were designed to not only fly you to the proverbial moon, but also to enable the flexibility to let you play among the stars by supporting a virtually limitless array of bandwidth-intensive applications and services through variable deployment scenarios. In other words--whereas legacy DSL platforms cannot get you to where you, as a wireline service provider, need to go--new multi-service platforms designed with video services in mind are essential to handle the bandwidth, capacity, management, and deployment challenges of this new emerging services domain.

Geoff Burke is Director of Video Solutions Marketing for Calix.

Visit Calix online.