2001: The dawning of the DOCSIS delivery platform?

Thu, 11/30/2000 - 7:00pm
Doug Jones, Chief Architect, YAS Corp.

Mankind has been in the world for 30,000 years. During the first 29,500 years, the agricultural age, we worked together and created a world of towns and cities of up to 100,000 people. During the last 500 years, we created the industrial age and developed cities of up to 15 million people and the workplace of the past century. In the last 15 years, we have started to build the information age. We have created a global economy where more than 1 billion people communicate together over the Internet.

Our vision and work over the next 10 years will continue to build on the Internet and create a broadband world. Experts predict that all people will be connected to vast amounts of content and media. This is a world in which video, data and voice will not only be truly integrated, but also a profitable business.

Cable, DSL, wireless, satellite and power line will all be competing to provide the best "first mile" connection from consumers to this broadband network.

Cable's access network offers the most aggregate "first mile" bandwidth into a subscriber's home. The cable access network is "channelized" into 6-MHz chunks of bandwidth as defined by the ITU J.83 standard. This channelization of the cable path allows flexible bandwidth allocation. It also allows different services to be placed on different channels. This is a unique property of cable networks, and can be referred to as frequency division multiplexing (FDM). That is, individual channels are at specific frequencies (e.g., the EIA channel plans).

Visit a large cable system headend and chances are video, data and voice services will be offered. But a closer look reveals that these services are offered as "parallel networks" that take advantage of the channelization of the cable access network. Not only do these services use separate channels on the cable plant, they also require separate sets of headend equipment. Cable operators have to procure separate systems, implement separate training programs, and run separate operations to offer these services. So, while cable operators offer voice, data, and video, they do so using three separate "silos" (see Figure 1). Although the huge bandwidth available on cable networks has allowed this strategy to work, the resulting amount of equipment and proliferation of complex technologies in the headend is growing to an unwieldy size.

The rapid increase in the amount of equipment needed to offer these services is a result of how the services evolved and what the suppliers offered as solutions. Voice and data have their own networks—the public switched telephone network (PSTN) and the Internet, respectively. Suppliers created proprietary product offerings to gain access to these existing networks over the cable access network. Back then, there was no universally accepted standard for two-way communication over the cable plant.

As a result, cable operators had to continually add more equipment to offer these services. All of this new equipment, each with its own complex technology, requires more capital expense, more operations expense, more training, more space, more air conditioning, etc. As will be introduced now, for competitive reasons, operators should consider a strategy to reduce the amount of equipment needed to offer voice, data and video services.

Figure 1: Offering data, voice and video services through three separate "silos."

Consumers are fickle. They want new services, and they care neither who provides them nor what access technology is used. In light of current and future competition, cable operators need to offer the most compelling services while maintaining a competitive cost structure. Cable cannot continue to deploy parallel networks and proprietary products for each new service. Parallel networks increase both capital and operations costs. Proprietary products limit both operators' flexibility and the control they maintain over their networks and services.

This article proposes the further convergence of DOCSIS- and MPEG-based services. DOCSIS and MPEG have shown that open standards facilitate both interoperable product development and competitive cost structures from multiple suppliers. Open standards also benefit suppliers by creating larger markets for their products. Cable operators should consider taking both the MPEG and DOCSIS efforts to the next level by specifying a new standards-based delivery platform that includes open interfaces to both the access cable network and subscriber equipment. This platform will facilitate both rapid application and service development and will lead to the first true generation of broadband services. To be complete, the proposal will also account for deployed legacy platforms and strategies to transition to the new platform.

This new platform is developing now, and uses DOCSIS as a transport mechanism. Note: the author does not advocate offering video services over DOCSIS. MPEG-2 transport is the most efficient method for delivering MPEG encoded digital video on a cable plant (see sidebar below.) Rather, this article advocates migrating set-top box (STB) control to DOCSIS.

The DOCSIS data transport standard is widely deployed in North America and the world. PacketCable, another effort underway at CableLabs, is reaching fruition in its goal of offering voice service over DOCSIS. The fact that data and voice services will share a common transport means they can use the same equipment in the headend. This saves on procurement, training and operations; hence, the attraction of convergence. The industry should not pursue convergence just for the sake of technology, but to position itself more competitively by lowering capital and operations expenses.

With voice and data services converging over DOCSIS, the next area for study is STB control. Here, STB control information is considered to be anything other than analog or digital video. This includes conditional access (CA) information, management and control messaging, application data, etc. In most North American deployments, STB control messages are carried on an out-of-band (OOB) channel. While these STBs all decode the same standard MPEG programming, the OOB transport channels from the two largest STB suppliers do not interoperate. The end result is that operators are locked into specific suppliers, which limits their flexibility and control over their own networks.

Cable has faced, and solved, similar situations with both data and voice products. Initially, there were proprietary offerings from suppliers. These are now converging over DOCSIS transport. Operators are gaining more experience with DOCSIS equipment and services every day. Technically, there is no reason why STB control messages can't also be carried using DOCSIS transport.

Another note about the competition. While DOCSIS has benefited the cable industry, it has also benefited its competitors. Modified DOCSIS technology is in development by both satellite and wireless providers for two-way access into subscriber homes. DSL suppliers are adopting DOCSIS-like processes to help them finally converge on an interoperable standard (see Figure 3).

Unlike voice and video, the challenge of migrating STB control will be the large embedded base of STBs. By the end of 2001, there will be an estimated 10 million deployed digital STBs using the legacy OOB channels. Any proposal to converge STB control over DOCSIS must address this legacy issue. In fact, there is such a solution. The legacy OOB channel will have to remain on the cable plant, but it is narrow and already accommodated in the spectrum allocation plan. The issue is with the legacy headend equipment that will need new interfaces to allow the control information to be dual-carried on a DOCSIS STB control channel. Because the legacy forward path QPSK channel is low bit rate, it can easily be dual-carried on a 6-MHz, 64-QAM forward path DOCSIS channel, with many megabits of additional capacity available to carry other interactive application data. The legacy set-top provisioning and conditional access systems (and hence the legacy STBs) can be used with both populations of STBs as long as the proper headend interfaces are made available. These legacy systems currently perform both device and service provisioning. The device provisioning will migrate to the already existing DOCSIS systems and the service provisioning systems will need to grow anyway to support the next generation of interactive services.

Figure 2: Converged DOCSIS delivery platform.

Another question is how to get DOCSIS into new STBs. This issue will solve itself. Most suppliers of advanced digital STBs have announced plans to include DOCSIS modems in their products to support interactive applications. Hence, these STBs will easily connect to the same headend equipment as the data and voice equipment. But even with a DOCSIS modem in the STB, the legacy STB suppliers continue to embed their OOB control mechanisms. Most advanced digital STBs that also include a DOCSIS modem will in fact contain two cable modems. One modem, generally either DAVIC (e.g., the Explorer 2000) or based on the ALOHA protocol (e.g., the DCT-2000), is used for STB control. The second modem, DOCSIS, is intended for interactive applications.

Figure 3: Point-to-multipoint access architectures that can use DOCSIS technology.

This business plan benefits the legacy STB suppliers by continuing to embed their OOB transport equipment. Additionally, the cost of "two modem" next-generation STBs are burdened with the additional modem. Going forward into a competitive environment, operators may lack both the flexibility and cost structure they might desire. Suppliers will argue that their legacy technologies work and are scalable. But a single-supplier solution lacks both the innovation and cost structure needed when facing competition for subscribers.

Using DOCSIS for STB control, with no legacy OOB channel, is in fact a reality. Pace Microelectronics has shipped more than 600,000 such devices and is continuing to advance the product line. Philips is rumored to be developing an "all DOCSIS" STB for another major European operator. The STB services an operator would expect, including conditional access and "carousel-type" services, are included. These STBs exist now and they do not use the legacy out-of-band QPSK carrier for STB control. Instead, these STBs use the same headend equipment, including the provisioning system, as both DOCSIS cable modems and PacketCable Media Terminal Adapters (MTAs).

The move toward retail also makes a strong case for migrating STB control to DOCSIS transport, with respect to both portability and cost. Most every North American headend will deploy DOCSIS transport for either data or PacketCable service, or both. In contrast, the Scientific-Atlanta and Motorola legacy STB transport protocols are each available in about half the headends, and they are generally mutually exclusive. Hence, a STB that communicates over DOCSIS will be inherently more portable. The Point of Deployment (POD) module, as currently defined, was necessary to provide portability because generally only one of the two non-interoperable STB transport protocols is implemented in a headend. By migrating to a common DOCSIS transport, which has no licensing fee, the cost of renewable security could be reduced.

The proposal is not all motherhood and apple pie, however. If DOCSIS is to be considered for STB control, the protocol needs to be reviewed for operation in one-way plant. Again, this is not as difficult as it sounds. For example, DAVIC is also designed for two-way communications, but DAVIC-based STBs continue to operate during a return path outage, albeit in a limited fashion. It's a similar engineering exercise to modify DOCSIS-based STBs to work in a one-way environment.

Also, there is a potential business issue with migrating STB control to DOCSIS transport. Operators are confronted with allowing customers to choose from a variety of Internet service providers (ISPs) for data service. What are the implications of including STB control over DOCSIS? Will operators be confronted with allowing subscribers to choose a provider for interactive applications? By including DOCSIS modems in set-tops, even if for just interactive applications and not STB control, the issue has been broached and will need to be addressed.

In summary, there are compelling reasons to converge STB control over DOCSIS. Data and voice are already there, and for competitive reasons, cable operators need to both remain efficient and regain control over their network and services. The biggest issues are the large deployments of set-tops using legacy OOB channels for STB control and the business issue of providing access to other service providers. The industry should research solutions to these issues.

The intent is to reduce the number of complex technologies in the headend. Each technology is burdened with capital cost (separate equipment) and operations cost (separate training, provisioning, etc.). Proprietary technologies are burdened with poor economics, low rates of innovation and less control for the operator. While cable has plenty of bandwidth to carry voice, data and video services, there are penalties in doing so with separate technologies. Moving forward into a competitive environment where other access providers will be offering voice, data and video services, cable should consider its competitive position. Cable has settled on DOCSIS as the method of choice for data and voice transport, and suppliers are including DOCSIS modems in set-tops. DOCSIS for STB control is the next step.




Should digital video be delivered over MPEG or over IP?

This is a technology choice. Broadly, MPEG-2 defines two technologies: First, a digital video encoding/compression technique; and second, a method for data transport over cable plant. MPEG-2 encoded digital video (ostensibly a form of data) can be delivered over either MPEG-2 transport or by encapsulating that digital video in IP and delivering it over DOCSIS transport. Both work.

That said, MPEG-2 transport is the most efficient method for delivery of MPEG-2 encoded video on a cable network. With MPEG-2 transport, there is less overhead—which translates to more "bits" being available for content delivery. Encapsulating digital video in IP is a reasonable way to deliver digital video if that content is sourced on the Internet (e.g., Real or QuickTime) or from a centralized data center that is connected to headends via a private intranet. Conversely, MPEG-2 transport is most easily used for content delivered off satellite or from a local server.

Both technologies work for delivering digital content and the proper choice depends on the context.


Share This Story

You may login with either your assigned username or your e-mail address.
The password field is case sensitive.