Converging Video, Voice, and High-Speed Data

Tue, 11/30/1999 - 7:00pm
Stephen D. Dukes, Vice President, Digital Broadband Technology, MediaOne Labs

With the advent of new services, the need for additional routing fabric grows beyond the capacity of a typical headend. The headend has traditionally consisted of equipment focused strictly on the broadcast of analog video and audio signals. However, with the introduction of new services and technologies such as high-speed data, circuit switched telephony, Internet telephony, and digital video and audio, most companies create a separate business entity without consideration of leveraging the same routing fabric of the network. Further, the floor space requirements have been or will be exhausted with the deployment of separate routing fabric to support each of these service types.

Figure I: Existing headend configuration

If the notion of a shared routing fabric is not considered, the migration of dedicated routing fabric can be costly and potentially fatal to a new business unit because it will be difficult to achieve the desired return on investment (ROI) within a targeted time frame. For cable operators to take advantage of the existing infrastructure, as they have in the past, the multi-system operator (MSO) must consider how to create a shared routing construct that enables the incremental addition of new applications and services. This must be accomplished without the use of dedicated routing fabric for each service or application.

If shared routing makes economic sense, what protocol and routing fabric should be used? The Data-Over-Cable Service Interface Specification (DOCSIS) is an appropriate starting point for the analysis of this question, because this is a standard that also uses Motion Picture Experts Group (MPEG) framing. MPEG is used for digital video and audio, and therefore, two major applications are considered directly.

The situation begs for a networking device that could carry traditional analog programs, digital video programs, expanded video services, voice services, high-speed data, and any other service that might arise in the future. The solution requires a common protocol and transport layer capable of efficiently delivering all of the various information types.

This article assesses possible network constructs for transitioning from the traditional cable network to a DOCSIS-based infrastructure that combines to support evolving services. The assessment will include quality of service, flexibility, scalability, and robustness or supportability of an overall cable infrastructure for data, voice, and video.

Opportunities and challenges

Cable companies recognize that digital networks afford major new revenue opportunities and increase the value proposition of a 6-MHz slot. Some operators see a lucrative market for video-on-demand, others recognize huge sales potential from entertainment gaming applications, and many expect that voice-over-IP will add to the corporate bottom line. But how does a cable business move forward into these new areas? What are the best methods for adding these services? What is the most cost-effective way to transition away from existing analog networks?

Traditionally, getting started with a new service meant installing equipment specific to the service offering and maintaining separate delivery systems, all within the spectrum of a cable system, e.g., 550 or 750 MHz. For example, an operator planning to move into digital video would typically install an MPEG or statistical multiplexer that would take the digitized video and output it in MPEG transport. But in addition to adding the multiplexer, the company also had to decide how to re-allocate the spectrum to accommodate the digital transport, which required sacrificing some number of analog channels to digital transmission. Without a comprehensive plan for allocation of programming within the spectrum, this can be a daunting task. In many cable systems, many analog programming contracts require access to the entire base of subscribers or do not provide for migration to digital. To satisfy the first requires a digital set-top terminal (STT) at each television or dual carriage. In many businesses where analog video accounts for 99 percent of the current revenues, making that decision is troublesome. It requires fixing bandwidth based on pre-supposed traffic patterns and speculative service adoption rates.

Fortunately, new technologies and communication protocol standards are making possible the migration to digital over a single transport infrastructure-one that handles both analog and digital traffic and can dynamically adjust to accommodate changing traffic loads. The goal is to migrate to an all-digital spectrum that dynamically allocates bandwidth to whatever service needs it. The enabling technology includes the implementation of a cable modem termination system (CMTS) that supports transmission of Internet Protocol (IP) packets over an MPEG transport construct. By evolving to this unified architecture, cable companies can make use of existing structures, minimize new equipment investments, and secure the technical flexibility to rapidly prototype new service offerings and optimally exploit market opportunities at a small incremental cost compared to the original separate mode systems.

The motivators: Multimedia applications

Multimedia applications incorporate some combination of audio, data, graphics, voice and video content. Increasingly, new services require interactive capabilities as well. While it is impossible to generalize or to determine which applications will ultimately be popular in any given market, some of the most promising areas and applications include:

  • Voice
  • Video entertainment
  • Internet information and entertainment
  • Enterprise communications (intranets, extranets).

To support the broadest set of applications-and offer the greatest revenue potential-a multimedia-enabled infrastructure must support interactive services and support audio, data, graphics, voice and video applications that converge to a single transport protocol.

These applications can be categorized into three major sub-groups, e.g., video, high-speed data and voice. The typical headend requires an analog headend, digital headend, circuit switch and CMTS to route or switch and combine for transport over the cable infrastructure. The cost of each system and the incremental cost to the network with low market penetration for several years increases the risk for the cable operator and is difficult to justify for new services and applications.

The introduction of new services is necessary for cable. However, many of these new services require incremental or dramatic changes to the infrastructure. A critical element for entry of new services is to insure these new services do not introduce network elements that can only be used by a single service. Routing is one such element that must be shared. Shared routing fabric enables multiple services to be offered over a single switch without extracting a heavy capital investment. This enables rapid prototyping of new services without large capital investments in routing fabric that can be stranded over time.

Prior to these new applications and services, the cable industry was focused on the challenges of broadcasting analog video and optimizing networks for the delivery of this service. This focus limited an operator's ability to pursue new opportunities. With the transition from broadcast analog video to multiple services, cable operators must optimize the network to support the delivery of an array of new services.

The notion of shared routing fabric can be applied to the rest of the infrastructure, as well.

The network requirements for multimedia

There are a number of technical challenges related to developing a network infrastructure that meet the business requirements of cable operators. The network structure must be robust to ensure consumer access, flexible enough to support the transmission requirements of a wide range of applications, and transparent to the consumer. To meet these objectives, an infrastructure has to adhere to the following requirements.

Interactivity, bi-directional functionality. Eventually, most multimedia applications will require interactive capabilities. These services require bi-directional functionality (downstream and upstream spectrum) and some form of routing, or routing and control. The switching or routing solution may be as simple as using multiplexers or a router, or possibly would consist of a more complex solution such as using fast packet.

Most cable systems will enable the return path. Downstream signals can exist in available spectrum on a sub-split system from 54 MHz to as high as 1 GHz, although 750 MHz is typically the upper boundary on most systems. The return path typically is limited to 5 to 42 MHz and is the limiting factor on the overall infrastructure. There are also mid-split and high-split implementations, but they require additional distribution equipment.

A passive coaxial network design would also provide additional upstream spectrum per subscriber, because the number of subscribers per node is smaller. This approach enables dynamic allocation of spectrum for upstream requirements. Because the amplifier defines how spectrum is allocated on the coaxial portion of the cable infrastructure, the elimination of the amplifier provides the means to manage bandwidth by transaction or demand basis, coupled with the routing component of the network.

Quality of Service. IP telephony applications require the capability of managing Quality of Service (QoS) to ensure that appropriate priorities and bandwidth are available for life-line services. DOCSIS 1.1 specifies the QoS needed to support voice-over-IP.

Security/privacy. The HFC topology is a shared infrastructure at the coaxial cable level. DOCSIS specifies baseline privacy plus, for security between the CMTS and the cable modem. Several hardware- or software-based approaches can be implemented. These include customer-owned encryption or a scheme that is inherent within the cable infrastructure such as conditional access for video content or movies. Additional security will be required, such as end-to-end security from the corporate office to the employee's home.

With the transport of voice, data, and video applications over this shared infrastructure, a more robust, renewable security system is necessary.

Bandwidth and throughput requirements. Most multimedia applications, today, are asymmetric in their bandwidth requirements in that the downstream bandwidth requirement typically exceeds the upstream requirement. Initially, a few channels may handle the demand, but as the demand increases a plan must be developed to increase upstream capacities. This can be accomplished by reducing node size or implementing more efficient compression algorithms.

Most current applications require average downstream throughput of 2 to 4 Mbps. On the return path, bandwidth requirements are much lower, ranging from 300 bps to 1.5 Mbps. Ethernet applications over cable requiring 10 Mbps throughput are available within a 6 MHz channel. A system operator may choose to limit access to bandwidth (e.g., 128 kbps to manage user expectations); however, ultimately, competitive pressures will require support of higher data rates.

Routing requirements. Many multimedia and interactive services require switching or routing functionality, even in a cable infrastructure that has traditionally been served well by a one-way broadcast. Approaches include circuit switching, packet switching, fast packet switching or routing (e.g., IP over DOCSIS). Only a routing-based construct can support multimedia applications in a single fabric.

Performance and reliability. Fiber optic facilities have yielded the largest improvement in performance and network reliability in cable infrastructures. Also, as fiber migrates further down into the infrastructure, improvements are realized. This is largely a result of the reduction in the number of active components-amplifiers and line extenders-in the network. Many existing network designs provide for no more than four amplifiers and two line extenders in a cascade to any one home. Newer designs seek to entirely eliminate the active components on the coaxial plant, and add a low noise amplifier at the residential unit.

Network management. Multimedia applications demand effective network management. Traditional approaches to maintenance are not reliable enough to support most multimedia applications. New approaches using neural networking techniques can detect a faulty amplifier before the failure actually occurs. For example, the practice of removing an amplifier from service to test its integrity cannot be tolerated. The passive coaxial network makes it easier for the cable operator to eliminate active components and to introduce an automated, dynamic approach to network management from a centralized facility such as the regional hub. This leads to quality improvements and operating cost economies.

Convergence around the CMTS will likely result in a paradigm shift to a unified management system across multiple services using an SNMP-based solution that offers fault detection and performance monitoring. In the future, network management will need to support other functions like relational databases.

Jitter and buffers. The delivery of MPEG-2 and MPEG-4 in a DOCSIS construct requires buffers at the end-points in the network to manage jitter introduced from the public switched telephone network (PSTN) and IP networks.

Mass storage. Transport exceeds the processing capabilities of most computer systems. For video and game applications, bandwidth can also be a scarce commodity requiring storage at the home. The variability between speed of transport and processing will necessitate some form of storage at national distributors, at regional hubs, headends, and at the home. In order to minimize the cost of mass storage systems, a hierarchical storage capability can be deployed to manage and optimize scarcity of bandwidth in the infrastructure. Several solutions exist between the STT and the consumer electronics device to manage traffic over the network.

Economics will generally determine where mass storage resides in the network. Operators may implement a hierarchical approach with distributed storage devices in the form of video servers located at the regional hub, the headend, the fiber hub or node, and even in the residence. The option of the consumer owning mass storage can reduce the MSO's capital investment in the network and increase flexibility for network operators.

Mass storage may include a standalone device resident between the digital STT and the consumer electronics platform-a television, VCR, or an embedded hard disk drive within the digital STT. The size of storage varies from approximately 14 to 28 GB, or 14 to 30 hours of programming.

Traffic. The network must support peak and non-peak, real-time and non-real-time traffic, without blocking or contention. The other challenge is to avoid significant overhead associated with the protocol usage for delivery of traffic over the network. With multiple services delivered over a single infrastructure, traffic management can occur at the CMTS or be controlled at the ingress points using RSVP resources for scheduling services and bandwidth.

Latency. Applications, such as games or video-on-demand, are sensitive to delays incurred over the network or in a server complex. Mapping MPEG-2 digital video into IP and IP into an MPEG packet may cause delay.

Network power. For lifeline telephony services, network power is essential. Whether power is derived from the network or from batteries in the home, a source of backup power is a requirement in the event of a commercial power outage.

Figure 2: Spectrum allocation for discrete implementation of voice, data, digital and analog video.
The existing network architecture

For some time, cable operators have been upgrading from the traditional coaxial-based systems to hybrid networks of fiber and coaxial infrastructure. Both designs deliver predominantly analog signals. Fiber offers improvements in system performance and reliability through its low-loss, increased-passband characteristics, and the significant reduction in the number of required active components. Overall, fiber greatly increases downstream passband capability. (Deployment of fiber further into the cable infrastructure results in smaller nodes and the elimination of trunk amplifiers. However, without the addition of new services, the generation of new revenues is not absolute.)

In coaxial networks, signals are collected from the program sources at the central headend and distributed to the home over coaxial cable. Trunk cables connect the headend to local distribution points where other cables branch out into the franchise area. Drop cables connect from the local distribution cables to subscriber homes. To overcome signal degradation, amplifiers are placed along the trunk and local distribution cables.

Figure 3: Migration to digital headend configuration.

In hybrid fiber/coax systems, fiber cables connect a headend to fiber hubs in a distributed topology in either a ring or star configuration serving 64 fiber nodes or 32,000 homes passed. These hubs feed fiber up to 64 fiber nodes that serve 500 homes passed. A fiber mini-node deployment enables fiber to 10 to 20 homes passed. The connections between the headend and the fiber hubs and the hub-to-hub interconnection uses a counter-rotating fiber ring that offers true physical route diversity.

The fiber node improves system performance, reliability and flexibility, and also reduces operating costs. In the case of the fiber mini-node, the fiber implementation eliminates active components (amplifiers and line extenders) in the network infrastructure, although it may require a low-noise amplifier at the home or in the NIU. The migration of fiber to the node can be optimized based on demand for spectrum and services with an associated revenue stream. Migration of fiber to the fiber mini-node may not offer a corresponding revenue stream until there is demand for the services.

Some of these new services will not be delivered over cable networks unless investments are made to modernize infrastructures in anticipation of actual demand.

At the headend, programs and other signals typically occupy a distinct frequency band, usually a 6-MHz slot.

As cable MSOs upgrade and introduce fiber, the networks are also expanding to carry other signals-e.g., high-speed data, voice-and the spectrum is extending into higher frequency ranges.

Figure 4: Spectrum allocation for migration of analog to digital video.

Converting or augmenting networks to carry digital signals offers great advantages for expanding network bandwidth: each 6-MHz slot can carry from eight to twelve digital program signals compared with a single analog signal. Because of the inhibiting costs of converting an entire cable network from analog to digital, cable operators typically add digital headends into the network and divide the spectrum to allocate passbands to the new digital programs. This scheme is still costly because it duplicates headends, and it requires that the traditional analog segment of the business sacrifice a portion of its programming spectrum. Until the new digital services actually generate healthy revenues, these overhead costs will severely impact the company's bottom line.

Data-Over-Cable Systems Interface Specifications

The underlining protocols that enable the convergence of these service types to a single fabric are DOCSIS and MPEG standards. DOCSIS 1.0 consists of a series of specifications and interfaces for supporting high-speed data over a cable infrastructure. The current DOCSIS international standard is published as ITU-T J.112 and provides a common physical layer definition for North America. The physical layer protocol is defined in ITU documents (ITU-T J.83 Annex B and modified ITU-T J.83 Annex B and J.112) which support a cell-based or IP-over-DOCSIS protocol.

Figure 5: Migration from circuit switch to VoIP configuration.

DOCSIS uses the MPEG packet as the cell construct for delivering IP-over-MPEG packets over a cable network infrastructure. The MPEG packet is a 188-byte packet with 4 bytes of header and 184 bytes of payload. For voice over IP and data, the packets are mapped into IP packets and carried over DOCSIS. The MPEG packet also serves as the packet construct for delivering digital video.

This approach offers a single transport protocol for delivering digital video, voice-over-IP, and high-speed data services. So why are there four separate systems-analog video, digital video, circuit switch, and a CMTS-when one will suffice? This represents a dramatic departure and potentially strands a significant amount of investment. Instead of new services struggling to survive with penetrations of three to five percent, consider the ability to rapidly prototype services without changing the network equipment. Instead of four different systems, companies could employ a single fabric based on DOCSIS.

A next-generation architecture

To introduce new services to an existing customer base, cable companies have been increasing the equipment and complexity of the existing network architectures while leveraging all of the existing broadcast, multicast or routing fabric. The results of this incremental approach include high start-up costs and increased management costs as a result of the resulting network complexity. It also introduces technical challenges and costs for in-home interfacing and service access.

Figure 6: Spectrum allocation for migration from circuit switch to VoIP.

Clearly, today's network infrastructure is capable of carrying a broad spectrum of combined signals. The situation begs for a networking device that could carry traditional analog programs, digital video programs, expanded video services such as VOD and conferencing, voice services, high-speed data, and any other service that might arise in the future. The solution requires a common protocol and transport layer capable of efficiently delivering all of the various information types.

The next-generation network achieves additional bandwidth through the conversion of the infrastructure from analog to digital and the use of IP mapped into DOCSIS. Digital enables the cable operator to map eight to 12 programming choices into a single 6-MHz slot, hence increasing the value of each 6-MHz slot. DOCSIS offers a method of more efficiently utilizing the bandwidth available in a 6 MHz slot.

Figure 7: Migration from digital headend to CMTS configuration.

With the introduction of the DOCSIS standards, it is possible to migrate the analog headend to a digital headend, map digital video into MPEG packets, transport IP in DOCSIS and circuit switched voice as voice over IP-all using DOCSIS as the underlying protocol. Voice-over-IP is accomplished through segmentation, fragmentation and concatenation of the MPEG packet and the addition of quality of service (QoS.)

Migration of fiber from a fiber node to a fiber mini-node serving areas of 10 to 20 homes passed can be more easily justified through the elimination of costly equipment at the fiber hub and fiber node and even at the mini-node-capabilities that an all-IP delivery protocol can offer. The adoption of digital and IP over DOCSIS enables the latter benefit to be achieved.

Consider that the DOCSIS standard for high-speed data transmission is already based on the MPEG transport where IP is mapped into DOCSIS or the MPEG packet itself. The same MPEG construct is also in use for digital video, and the foundation for voice over IP, that enables efficient delivery of multiple services. If all analog video were converted to digital and then transported in IP, one broadcast/routing fabric could be used for all traffic.

By transitioning all traffic to DOCSIS, the headend configuration can be greatly simplified. Instead of the existing analog headend, digital headend, circuit switcher and CMTS, the headend architecture could be consolidated to a single CMTS that can be used to deliver interleaved video, voice and data services to the common, existing broadcast/routing fabric. While the CMTS currently supports access to a single 6 MHz, ultimately the CMTS will support multiple 6-MHz slots.

A common protocol and transport construct would greatly affect spectrum allocation. Currently, existing analog programming and new services are both restricted to a fixed number of 6-MHz slots. Even with a move to digital services, and the resulting increase to as many as 10 programming choices in a single 6-MHz slot, peak traffic in any one service area creates the risk of bottlenecks.

With the DOCSIS construct and the use of a single infrastructure for all traffic, the spectrum can be handled in a different fashion. At any point in time, 6-MHz slots can be allocated where needed with no fixed frequency ranges or limits on the number of slots for any particular service. The full spectrum can be made available in a manner that complements the revenue streams, and the inefficiencies of dedicated or fixed programming can be eliminated.

Many companies are emerging in the consumer electronics market, and the resulting offerings of in-home solutions are expected to increase dramatically in coming years. A common networking infrastructure encourages the development of standard in-home interfacing options, and will accelerate the proliferation of affordable, intelligent devices for the consumer. The cable modem in this scenario will serve as the network interface unit to the home. The NIU will serve as the interface to multiple cable modems and digital STT or set-back devices within the home or embedded in the consumer electronics platform or personal computer made possible by the CMTS-based headend architecture.

By eliminating the need for numerous and complex headend configurations, companies reduce the costs of delivering data to the infrastructure. CMTS products can be located where needed, simplifying the management of the overall infrastructure. Many of the functions that are currently duplicated at each analog/digital headend can then be moved up a level in the network hierarchy with highly centralized regional functions. A single, centralized back office configuration can manage bigger regions. This next-generation back-office will replace numerous, distributed headend systems with one implementation that can support:

  • Conditional access systems (CAS)
  • CMTS
  • Circuit switches (existing phone system access solutions until the public networks transition to IP)
  • Game modem pools
  • Billing systems
  • MPEG or statistical multiplexers
  • VOD servers
  • Analog video (off-the-air or from other sources)
  • Caching
  • File servers
  • Fast Ethernet Switches
  • Gatekeeper functionality
  • Gateways
  • Advertisement insertion

Just as many functions can be centralized at a higher level for reduced equipment costs and simplified management, the new IP/DOCSIS infrastructure also enables the placement of intelligence further downstream. For example, in-home intelligent devices could be designed that could control the downloading of advertising tailored to the home's buying history. The CMTS and cable modem connection supports a variety of devices with intelligence placed where it best accomplishes revenue goals and consumer satisfaction.

Previous efforts have focused on MPEG mapped into IP packets. These efforts have failed to gain the momentum needed to serve as a single protocol for transporting voice, data and video applications. With the advent of DOCSIS, IP over DOCSIS has enabled not only high-speed data, but voice-over-IP and digital video over DOCSIS.

The key enabling hardware elements are the CMTS and the cable modem, combined with the use of routers deployed at the hub to route traffic to the various fiber nodes. These system components are based on standards that serve as the platform for DOCSIS.

Benefits of the IP-MPEG solution

The collapse of the traditional headend architecture into a distributed network of CMTS devices results in far-reaching benefits for the individual cable providers and the industry as a whole.

Today, the logistics and start-up costs for introducing a new service demand costly planning and implementations since each new service can translate into additional infrastructure capital expenditures. With a move to IP over DOCSIS across the entire network infrastructure, cable operators can reduce the incremental cost to introduce a new service using the same broadcast/routing fabric and the same CMTS delivery system as existing services. More services can be bundled, improving the company's competitive advantage. Maintenance costs are also significantly lowered with the less-complex architecture and the many centralized functions that were previously duplicated at each headend.

Figure 8: Integrated CMTS supporting voice, data and video.

A common infrastructure capable of transmitting any data type gives cable companies the opportunity to inexpensively and rapidly prototype new services or deploy new services on a limited or test basis. New services do not require any changes to the headend equipment, network equipment, or NIU. Companies can quickly introduce the new services to customers. Introduction cycles are also shortened because a new service does not require the design or availability of a new delivery system (customized servers or devices).

The CMTS/CM network and the IP over MPEG transport construct defined by DOCSIS 1.0 and 1.1 give cable MSOs the flexibility to maximally place intelligence to meet business goals while ensuring customer satisfaction. With game services, for example, the game servers can be centralized so that costly duplication of servers can be avoided while still covering a large region. The IP over DOCSIS ITU-T J.112 standards based on MPEG transport promise to efficiently support a broad range of application architectures, both centralized and distributed.

By supporting dynamic allocation of the spectrum based on real-time traffic requirements, the IP over MPEG infrastructure makes optimal use of bandwidth. With more bandwidth available, cable companies can not only introduce more services and reach more homes with any given infrastructure, but they can guarantee various levels of service and evolve to a pricing structure based on QoS.

Convergence of the various service types creates simplification to the network infrastructure and to back office functions such as billing systems, network management, QoS, signaling, security, scalability, routing fabric and bandwidth management. For example, with four separate service types, integration of billing is far more complex and typically results in a single, fixed billing scheme to accommodate the combination of services. With a single network fabric, a cable operator will eventually be capable of offering both fixed and variable rate billing based on usage. Similar benefits exist for each of the other functions mentioned above.

A single network fabric also enables further degrees of integration at the interfaces. For example, the WAN interface and CMTS can be integrated into a single edge device with a common interface like a bus or crossbar switch.

The DOCSIS-based network allows large and small cable operators, with small and large systems alike, to offer a common set of services. The challenge with many of these new services and technologies has been that many small and medium operators cannot offer the full suite of services that many of the large cable operators currently deploy. A single fabric provides a scalable approach for offering a full complement of services without an investment in multiple discrete systems.

Migration example

To understand migration to the next-generation architecture just discussed, consider as an example a headend complex that includes analog and digital headend equipment, a circuit switch for wired telephony and a CMTS for high-speed data, as illustrated in Figure 1. In this typical headend configuration, the array equipment adds complexity and occupies floor space. These systems support voice, video applications (video-on-demand, video conferencing), high-speed data, advertisement delivery and interactive games.

Spectrum is allocated from 54 to 650 MHz for analog video, 650 to 738 MHz for digital video, 738 to 744 MHz for high-speed data, and 744 to 750 MHz for circuit switched voice, with 5 to 42 MHz dedicated for upstream or return path as shown in Figure 2.

Analog to digital. The first phase of migration for this system includes deployment of digital STTs and conversion of all analog spectrum to digital. This first level of integration converts the analog headend to a digital headend, as illustrated in Figure 3. The circuit switch and CMTS can remain in use for support of wired telephony and high-speed data.

The spectrum previously used for analog is converted to digital, increasing the number of programming choices available, as shown in Figure 4. The digital programming resides in the 54 to 738 MHz of spectrum. The spectrum for wired telephony and high-speed data remains the same.

Integration of data and voice: DOCSIS. The next stage of integration leverages the fact that voice-over-IP is based on the DOCSIS 1.1 protocol and PacketCable. This enables the integration of high-speed data and IP voice traffic in a single fabric-the CMTS. Figure 5 illustrates this integration. The circuit switch may be used for access to the public switched telephone network (PSTN) or a Tandem. Optionally, the circuit switch may be eliminated completely.

Integrating the high-speed data and voice-over-IP enables the use of a CMTS to support both applications. The spectrum allocated for each application can be used separately or combined, as shown in Figure 6. The allocation of 738 to 750 MHz remains the same or could be reduced to a single 6-MHz slot, initially.

A single transmission protocol: DOCSIS. The final stage of integration is based on the notion that digital video and high-speed data are integrated into the same underlying transport protocol-MPEG packets—as illustrated in Figure 7. This construct enables a single fabric-the CMTS (as illustrated in Figure 8)-for routing voice, data, and video at the headend and throughout the network in a single transmission protocol. This integration results in a spectrum allocation as shown in Figure 9. The use of a single transmission protocol also reduces complexity at the fiber hub because the only functionality at the hub is a router.

The use of the CMTS as the primary network element enables the cable operator to manage a wide range of services within the network without high incremental costs. Advanced services can co-exist with core services without a high incremental cost. With the growth of services, such as high-speed data or IP voice, the operator can use the same fabric to incrementally expand these new services without the risk of stranding capital investments. The same human resource used to manage the video can support the data or the voice, since the CMTS is the single element that aggregates all the services at the headend.

Initially, this is not practical. The cost of deploying cable modems as NIUs in every home along with a digital setback device for every television is too high. The alternative is to migrate to the DOCSIS platform by allocating a portion of the spectrum and/or applications to this platform, and, over time, migrate to an all-digital DOCSIS platform.

Companies can allocate 650 to 750 MHz to the DOCSIS platform, including the conversion of circuit-switched voice to IP telephony using DOCSIS 1.1, high-speed data, and digital video-on-demand. Video-on-demand service requires a digital STT. The content for video-on-demand consists of film and NTSC video. Games and other latency-sensitive programming can be added in future migrations. This gradual migration enables the cable operator to carry analog and digital programming from the outset and, eventually, migrate to a DOCSIS platform that occupies the entire spectrum.

This example applies to a 750 MHz system; however, the same migration can be applied for a 450 or 550 MHz system.

Figure 9: Spectrum allocation for integrated voice, data and video.

Changes to the network. Back office. Simplification to a CMTS results in a new back office, and enables the cable operator to move this functionality one level higher in the hierarchy to a regional headend facility.

CMTS. The use of a CMTS allows the cable MSO to manage various versions of DOCSIS with a common backplane and separate line cards. As IP telephony roles evolve, a CMTS can support DOCSIS 1.0 and 1.1 without the need for separate systems. Just as the transmission of multiple programs over a single coaxial cable, the CMTS offers the cable operator the capability of supporting multiple service types through a single routing fabric.

The CMTS will likely evolve to an edge device that supports the DOCSIS high-speed interface on the WAN side and the traditional CMTS function on the local distribution side interconnected by a cross-bar switch type functionality.

Hub. The functions of an HDT, HSD, PEG, analog and digital headend, etc., in the hub are simplified because the only functionality that resides at the hub consists of a router, some storage, and possibly local advertisements for insertion. The remaining functionality is migrated to the headend or higher in the network infrastructure.

NIU. The NIU effectively is the cable modem. It reduces complexity over the network. The NIU with an integrated cable modem and 1394 interface offers the capability of supporting consumer electronics and computer devices with embedded digital STTs and cable modems to exist in a more simplified format. The digital STT may become less complex with more of the functionality embedded with the television platform.

Setback device. The output of a cable modem-based NIU is probably a 1394 interface that supports bandwidth for voice, data, and video applications. Because the cable modem does not perform the functions of an STT, a setback device is required at each television or embedded within every consumer electronics device. The digital setback device will likely need to have buffers to support jitter and time stamp derived in an IP network.


As cable companies strive to reach more homes and attract more customers, the cable MSO that can offer more services while reducing the costs of expanding its subscriber base will have the competitive advantage. By evolving existing network infrastructures to a common IP over DOCSIS protocol, companies can more rapidly and economically introduce new services, while preserving investments in existing equipment and the physical wiring connecting to subscriber homes. DOCSIS and MPEG standards can address all of the information transmission requirements today and in the future, and will enhance the overall revenue-generating opportunities in the cable industry. This approach optimizes the cable infrastructure for new services and allows the efficient, dynamic allocation of bandwidth across services.



Share This Story

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