March 17, 2011, 5:35 a.m.
posted by ksch
Cable TV Networks
Cable operators, also referred to as multiple-system operators (MSOs), are in major competition with telcos because the two industries are increasingly resembling each other, offering the same range of services and fighting for the same customers: Cable providers have gone into the business of supporting data and voice communications (e.g., Internet access and VoIP), while telcos have begun to provide TV and interactive services (e.g., IPTV and VOD). Cable TV networks, which are most commonly deployed as hybrid fiber coax (HFC) arrangements, support a wide range of services, including traditional circuit-switched telephony and VoIP, Internet access, broadcast video, VOD, and interactive broadband services. HFC involves the use of fiber in the backbone and in the access network. The fiber termination point (i.e., the neighborhood node) can support anywhere from 200 to 2,000 homes, with 200 to 500 homes being the norm. From that neighborhood node, coax (normally 750MHz or 1,000MHz) is run to the home, in a two-way subsplit system. (When two cables are not used but there is a need for simulation of a dual-cable system, bandwidth on a single cable can be split up, with one portion representing one cable, and the other portion representing the second cable. Splitting the frequencies so that the lower frequencies are used for one purpose and the higher for another is called a subsplit system.)
In countries where there is a history of cable TV, the cable plants have traditionally been one-way analog networks, sending just downstream; after all, what more was required for the conventional broadcast of live television and video programming? However, to handle Internet access, voice communications, or any other interactive services, a two-way infrastructure is required, so the operators had to upgrade their networks. Over the past few years, the upgrade to digital two-way systems has been occurring in countries with existing cable infrastructures. In other parts of the world, where there had been no MSOs, new systems are being built with digital two-way capabilities, recognizing the need to support today's interactive services environment.
A number of elements are involved with a cable TVbased network, including the physical transport infrastructure, generally known as the HFC architecture; cable modems, which are largely standardized under the CableLabs DOCSIS specifications; and cable modem controllers, which are known as cable modem termination systems (CMTSs). The following sections discuss these elements.
Cable operators' network infrastructures are based on the HFC architecture, which is in essence a community LAN that uses a bus topology, meaning that it's a shared-access architecture. (Chapter 6, "Local Area Networking," describes bus and other LAN topologies.) Figure shows the topology of a cable TV operator's HFC network. On the left side are the headends, from which the information is being broadcast. Those headends are tied together, generally with fiber, referred to as fiber in the backbone. By moving to a fiber-based backbone, the cable TV operators have also made improvements to the performance of their networks as well as the costs associated with operating and maintaining them. The backbones, then, feed into the neighborhood nodes, or the optical nodes (i.e., the access points), and from there coax goes out to the homes. You can see in Figure that HFC is a shared infrastructure; this is one of the drawbacks of the HFC architecture.
3. The topology of an HFC network
The HFC architecture uses Frequency Division Multiplexing (FDM) to derive individual channels, some of which are dedicated to the support of telephony services, others of which are reserved for analog TV, and still others of which are reserved for DTV and future interactive broadband services. This multiple-access coax system represents a bit of a hostile environment: The points where the coax connects into set-top boxes or cable-ready TV sets tend to collect noise, so the cable picks up extraneous noise from vacuum cleaners or hair dryers. If every household on a network is running a hair dryer at 6:30 AM, the upstream paths are subjected to this noise, and hence there will be some performance degradations. Extra signal processing must therefore be added to overcome the impairment in the return channel.
The major concerns with HFC include security, privacy, reliability, and return-path issues, particularly in support of telephony. With twisted-pair, we're used to having a private line to the local exchange, but that is not the case in using a shared coax system. Also, HFC faces bandwidth constraints: Given that cable modems have caught on and there are increasingly more subscribers to such services, more homes within a neighborhood are making use of their Internet access channel, and everyone's downloads and bandwidth are minimized. Subdividing the nodes can help alleviate bandwidth constraints, and it can also help reduce ingress noise. If the service provider continues to subnet, it can keep performance high for the subscribers.
Cable Modems and CMTSs
A cable modem is needed in order to support high-speed data access over HFC, using the cable TV infrastructure. Cable modems function like special-purpose routers, linking the cable network's Layer 3 to another network or device. Generally, this requires an external box with cable and Ethernet connections. Figure illustrates cable modem connectivity. On the left side of the figure is an individual neighborhood with users attaching to the shared coax system via their cable modems. These various coax trunks then come into the headend facility, where they terminate on a CMTS. The CMTSs are linked by a common Ethernet hub, which in turn feeds into the IP router, which determines the optimum path to take over an optical backbone onto the ISP.
4. Cable modems: LAN-oriented connectivity
CMTS functions include providing QoS, allocating bandwidth, classifying packets, policing packets for Type of Service (ToS) fields, adjusting the ToS fields as needed, performing traffic shaping, forwarding packets, converting and classifying QoS parameters, handling signaling and reservation of backbone QoS, and recording call resource usage.
Cable modems provide downstream data rates of up to 36Mbps, and the downstream rates are generally supported in the frequency band 42MHz to 750MHz. The downstream channel depends on the QAM technique used because that is what gives it the most bits per second and, hence, the fastest data rates downstream, where rapid downloads are important. Upstream data rates range from 5Mbps to 30Mbps, and the upstream direction operates in the range of 5MHz to 42MHz. As mentioned earlier, this portion of the frequency band is especially subject to noise interference, so it requires modulation techniques such as Quadrature Phase-Shift Keying (QPSK) and 16-QAM, which transport fewer bits per second than other techniques but also provide better noise resistance.
Cable Modem Standards
Many standards deal with cable modems. CableLabs (www.cablelabs.org) is an industry leader in creating cable modem standards for the cable TV industry. CableLabs has defined the following standards:
The Euro-DOCSIS standard was created based on the U.S. DOCSIS standard, with the goal of ensuring correct and optimal performance of Euro-DOCSIS modems and CMTSs in European networks as well as being fully compliant with the European Digital Video Broadcasting (DVB) standard in the downstream. tComLabs (www.tcomlabs.com) developed the Euro-DOCSIS annex of the DOCSIS specification in 1999, with the cooperation of equipment vendors and cable operators. The European cable community agreed on final specifications in early 2000.
ETSI has approved both DOCSIS and the Euro-DOCSIS annex as specifications. The primary difference between DOCSIS and Euro-DOCSIS is that DOCSIS employs the 6MHz channel NTSC standard used in North America, whereas Euro-DOCSIS relies on the 8MHz PAL channel spacing used in Europe. Also, Euro-DOCSIS takes advantage of a higher capacity in the upstream band, ranging from 5MHz to 65MHz, versus the North American version, which uses 5MHz to 42MHz. Japan employs other variants of DOCSIS.
For several years, cable TV operators have been migrating from their traditional core business of TV (or entertainment) programming to the role of a full-service provider, offering not just video but voice and data services as well. DOCSIS has been instrumental in this shift. The main objective of DOCSIS is to provide uniform specifications to ensure compatibility with various cable operators' infrastructures and to allow cable modems to be purchased from various retail outlets.
There are two main components to the DOCSIS architecture: the cable modem, which is located at the customer premises, and the CMTS, which is located at the cable operator's headend. The customer's equipment, such as a PC, is first connected to the cable modem, which is then connected through the HFC network to the CMTS. The CMTS basically performs the same function as the DSLAM does for xDSL: hosting downstream and upstream ports and routing traffic between the HFC network and the Internet. The cable operator also uses the CMTS to configure each customer's cable modem, adjusting for different line conditions based on the customer's service requirements.
With the DOCSIS and Euro-DOCSIS standards, the downstream channel occupies the capacity of a single TV transmission channel in the cable operator's channel offerings: In the United States, it is a 6MHz downstream channel, and in the European annex to the standard, it is an 8MHz channel. In general, several hundred users can share the downstream channel and one or more upstream channels. The digital set-top box uses MPEG-2 transport stream modulation, based on either 64-QAM or 256-QAM, providing up to 40Mbps downstream. The upstream can use QPSK, 16-QAM, or 64-QAM, depending on the generation of DOCSIS standard. (See Chapter 5, "Data Communications Basics," for information on modulation schemes such as QPSK and QAM.) DOCSIS also involves an Ethernet connection to the PC, so data is transferred by using TCP/IP encapsulated in Ethernet frames between the cable modem and headend. DOCSIS includes a baseline privacy specification that relies on the use of both the 40- and 56-bit versions of DES. (See Chapter 9, "IP Services," for more information on security.)
At this point, there are several generations of DOCSIS standards, including DOCSIS 1.0, 1.1, 2.0, and 3.0. The ITU has also adopted two of the DOCSIS variants as international standards: DOCSIS 1.1 was ratified as ITU-T Recommendation J.112, and DOCSIS 2.0 was ratified as ITU-T Recommendation J.122. DOCSIS 2.0 is backward compatible with DOCSIS 1.1.
DOCSIS 1.0 enables the cable TV industry to deliver high-speed data using cable modems. As with all the other DOCSIS standards, the main service with DOCSIS 1.0 is two-way access to the Internet.
With DOCSIS 1.0, the upstream rate is up to 10Mbps over a 3.2MHz channel. For downstream data, the modulation technique specified is either 64-QAM or 256-QAM, and upstream it is either QPSK or 16-QAM. It is possible to deploy DOCSIS 1.0 on a one-way HFC network by implementing the return path over traditional phone lines.
DOCSIS 1.1 was created to address the cable industry's desire to offer VoIP services. DOCSIS 1.1 includes key network technologies, including dynamic QoS, which is very important to VoIP, packet fragmentation, and enhanced security. (QoS is discussed in Chapter 10.) By supporting QoS, DOCSIS 1.1 allows for the provisioning of VoIP and interactive gaming. QoS also enables tier-based services such as higher speeds to heavy users. DOCSIS 1.1 also offers improved security and privacy, both of which are necessary to support voice services. For downstream data, the modulation technique specified is either 64-QAM or 256-QAM, and upstream it is either QPSK or 16-QAM. Along with cable modems, DOCSIS 1.1 also supports VoIP phones and residential gateways.
To deploy DOCSIS 1.1 (or higher), the cable operator must have in place a two-way HFC network that supports a return path for the upstream traffic. The DOCSIS 1.1 standard addresses real-time applications such as IP telephony. The key issues in cable-based IP telephony include voice quality and how to guarantee it in terms of latency, fidelity, jitter, packet loss, and reliability at the customer end. Other issues are legacy signaling support, data security, scalability, and feature deployment at the service provider's end. Finally, there are a number of provider-specific issues, such as implementation of systems for PSTN gateways and gatekeepers, provisioning, billing, and network maintenance.
DOCSIS 1.1 enables time-sensitive voice and multimedia packets to share in HFC networks with time-insensitive pure data packets. DOCSIS 1.1 enables a node to recognize a nondata packet and switch to it instantaneously from whatever data packet it is working on. It requires a CMTS at the edge of the cable access network and a DOCSIS 1.1compliant cable modem at the customer premises. Edge cable CMTSs need the intelligence to isolate traffic flows and apply policy-based QoS treatments in real-time. Traffic flows need to be isolated by service provider, application, and subscriber so that during times of congestion, flows within the service-level agreement (SLA) are maintained and flows that exceed the SLA are discarded first. Operators then map the DOCSIS-based flows to IP specifications such as DiffServ and MPLS, which are discussed in Chapter 10, to manage the handoff to the core network.
Figure shows an example of using DOCSIS 1.1 for cable-based IP telephony. Considerations for cable-based IP telephony include technical architecture, achieving PSTN-level reliability (i.e., five-nines reliability, or 99.999%), being able to accommodate the same PSTN-level feature sets, and regulatory issues. Operators face challenges such as how to provide detailed, sophisticated, end-to-end SLAs; how to adjust to the need to do maintenance, which will become more critical; and how to evolve from being broadband video providers to being mission-critical service providers. Developments in the cable-based IP telephony environment are expected in the next couple years.
5. Cable-based IP telephony
DOCSIS 2.0 and 2.x
DOCSIS 2.0 increases downstream speed to 40Mbps and upstream throughput to 30Mbps over channels as wide as 6.4MHz, resulting in an increase in the capacity to deliver high-speed data. DOCSIS 2.0 also supports symmetric services, which means it can serve business customers more adequately. Along with broadband Internet access, VoIP, and tiered services, DOCSIS 2.0 also supports commercial services and videoconferencing. In addition to cable modems, VoIP phones, and residential gateways, it also supports video phones.
For downstream data, the modulation technique specified is either 64-QAM or 256-QAM, and upstream it is either QPSK, 16-QAM, or 64-QAM. CableLabs and its members are working on a 2.x version of DOCSIS that will add features such as roaming, committed data rates, Media Access Control (MAC) layer improvements, and better commercial service capabilities. DOCSIS 2.x also adds support for mobile devices.
More than two years after CableLabs officially released the DOCSIS 2.0 specification, more than 60 cable modems are now DOCSIS 2.0 certified, and the technology is a must-have on modem supply contracts with major operators. However, only a handful of operators have actually deployed DOCSIS 2.0 in their networks. It is only when the cable operators are ready to offer VoIP and commercial services on a large scale that they will begin using DOCSIS 2.0. Although most operators will build it into their hardware, it will be the servicesfrom IP voice to multimedia gamingthat will drive MSOs' decisions to implement DOCSIS 2.0.
DOCSIS 3.0, currently under development, will increase the capacity up to a minimum of 160Mbps downstream to customers and a minimum of 120Mbps upstream. In DOCSIS 3.0, several downstream and several upstream channels can be bonded together to multiply the bandwidth delivered to each customer.
DOCSIS 3.0 will most likely require a new chip at the modem and at the CMTS. However, DOCSIS 3.0 will be worth the effort because it will give operators enough bandwidth to offer a wider range of IP-based offerings, including entertainment-quality media services. (Figure compares the amount of bandwidth each DOCSIS version provides.) DOCSIS 3.0 adds entertainment video on top of the services supported by DOCSIS 2.0 and 2.x, and it also adds support for IP set-top boxes to the list of equipment supported by the earlier DOCSIS versions.
The cable industry is applying DOCSIS to many services and applications beyond just Internet access, including new applications to control the communications pathway for digital set-top boxes and to monitor the health of the cable plant, including power supplies and amplifiers.
DOCSIS Set-top Gateway (DSG) is an extension to the DOCSIS standards that gives operators a standard method to deliver out-of-band data, such as channel lineups and program guides, and more advanced streaming applications via a DOCSIS channel to the digital cable set-top box.
The DSG network consists of three main elements: the DSG gateway, which is responsible for generating the stream of out-of-band data; the DSG agent, which is the same as a CMTS and is responsible for forwarding the out-of-band data as well as publishing the out-of-band directory; and the set-top box. The set-top box contains three subcomponents: the DSG-capable cable modem, also called the DSGeCM (embedded cable modem); the actual data consumer, called the DSG client; and the DSG client controller, which configures and controls the DSGeCM. The DSG specification applies equally to all versions of DOCSIS.
CableLabs created PacketCable (www.packetcable.com) to define standards for the cable TV industry. The PacketCable initiative aims to develop interoperable interface specifications for two-way cable networks in order to deliver advanced real-time multimedia services. PacketCable interconnects three networks, including the HFC access network, the PSTN, and IP networks. PacketCable networks, built on top of DOCSIS 1.1 or higher, use IP to enable various multimedia services, including IP telephony, videoconferencing, multiparty game playing, and other multimedia applications. By using a single high-speed, QoS-enabled broadband architecture, a DOCSIS 1.1 or 2.0 network with PacketCable extensions allows cable operators to economically and efficiently deliver data and voice traffic.
The PacketCable architecture relies on several key protocols. First is DOCSIS, which is the standard for data over cable and details the radio frequency (RF) band. For media transfer, Real-Time Transport Protocol (RTP) and Real-Time Transport Control Protocol (RTCP) are required. The signaling protocols include PacketCable Trunking Gateway Protocol (TGCP; a PSTN gateway call signaling protocol specification), which is an MGCP extension for media gateways, and the Network-Based Call Signaling (NCS) protocol specification, which is an MGCP extension for analog residential media gateways that details VoIP signaling. (RTP, RTCP, and MGCP are discussed in Chapter 9.) Finally, Common Open Policy Services (COPS) is used for controlling access. (COPS is discussed in Chapter 10.)
PacketCable is known internationally as IP Cablecom. PacketCable documents have been approved by the Society of Cable Telecommunications Engineers (www.scte.org), which ANSI recognizes as a standard-setting body for cable in North America, and have been approved by the ITU for adoption as worldwide standards. The cable industry is pursuing global standardization, with the objective of achieving worldwide interoperability of services and equipment, vendor independence, ease of interconnection with other operators, and reduced cost through scale of economies. The ITU has approved a suite of CableLabs PacketCable specifications as standards for the international version of services, including VoIP.
As discussed in the following sections, three versions of the PacketCable standard are defined today: 1.0, 1.5, and 2.0.
The first service defined for the PacketCable platform is VoIP. The CableLabs PacketCable 1.0 specification deals with transmitting multifeatured IP phone calls over HFC, and it allows four independent IP voice channels through a single cable modem. PacketCable 1.0 consists of 11 specifications and 6 technical reports that define the call signaling, QoS, codec, client provisioning, billing event message collection, PSTN interconnection, and security interfaces necessary to implement a single-zone PacketCable solution for residential IP voice services.
The core set of PacketCable 1.0 specifications essentially describes how to move the basic functions generally found on traditional Class 5 local exchanges onto a distributed architecture consisting of several general-purpose servers, leading to a low-cost, highly flexible, and scalable solution. PacketCable multimedia defines a generic architecture where application managers request QoS on behalf of a client and where policy servers authorize and commit these QoS requests.
PacketCable 1.5 supersedes previous versions (1.1, 1.2, and 1.3) and consists of additional capabilities. It consists of 21 specifications and 1 technical report that define the call signaling, QoS, codec, client provisioning, billing, message collection, PSTN interconnection, and security interfaces necessary to implement a single-zone or multizone PacketCable solution for residential IP voice services.
PacketCable 2.0 will replace MGCP with SIP. (The MGCP and SIP protocols are discussed in Chapter 9.)
Digital cable TV devices present another exciting area to watch in the coming years. The goal of the CableLabs OpenCable program (www.opencable.org) is to publish specifications that define digital cable network interfaces, as well as the nature of next-generation cable set-top boxes. CableLabs started the OpenCable initiative in 1997, with the goal of helping the cable industry deploy interactive services over cable, creating a common standard for digital cable TV within the United States, and promoting competition among licensed device manufacturers.
OpenCable has two key components: a hardware specification and a software specification. The hardware specification describes a receiver that ensures interoperability and can be sold at retail. The software specification, called the OpenCable Applications Platform (OCAP), creates a common platform on which interactive services can be deployed. The following sections describe these two components.
The CableLabs cable modem standard MCNS will be used with OpenCable set-top boxes, with advanced digital video compression circuitry to create terminals capable of supporting next-generation video and the entire range of current and future Internet and Web-based applications. The OpenCable effort is seen as the linchpin of the cable industry's digital future. It is independent of the processor and operating system. Compliant set-top boxes must allow both high- and low-speed bidirectional Internet service for both Internet and TV applications, and computer applications must be provided to both the television and the desktop computer through cable.
Characteristics of OpenCable digital set-top boxes will include expanded memory, powerful graphics engines, and support for one-way broadcasts (e.g., near-VOD, Web browsing, Internet e-mail) as well as two-way interactive services (e.g., Internet access via TV, high-definition video programming). Besides defining next-generation digital consumer devices, OpenCable also aims to encourage supplier competition and to create a retail hardware platform.
OCAP is intended to enable the developers of interactive television services and applications to design such products so that they will run successfully on any cable TV system in North America, independent of set-top or television receiver, hardware, or operating system software choices.
OCAP consists of a set of technical specifications created by CableLabs and endorsed by the Society of Cable Telecommunications Engineers (www.scte.org) and other industry groups. The OCAP 1.0 specification was released in December 2001 and was followed by OCAP 2.0 in April 2002. The specifications include two main sets of software: middleware and applications software/content authoring tools. The majority of the software is based on Sun Microsystems Java, which is already used by many developers to create content for PCs, the Web, gaming devices, and TV. OpenCable's objective is to put OCAP middleware into all sorts of intelligent devices and then use OCAP authoring tools to create interactive content to run on those devices. This will allow manufacturers and retail distributors of set-top boxes, television receivers, and other appliances to build and sell to consumers interesting and highly functional next-generation digital devices aimed at supporting a host of new and exciting interactive services delivered by cable operators. Simply put, OCAP creates the opportunity to establish a standardized platform, national and international, to launch a myriad of interactive services over a wide variety of digital devices.
Supporters of OCAP believe that it will do for TV in the United States what DOCSIS has done for high-speed Internet access and what DVBMultimedia Home Platform (DVB-MHP) is doing for TV in Europe. (DVB is discussed in Chapter 10.) In fact, much of the OCAP specification is based on or is compatible with MHP. In addition, South Korea has adopted OCAP to introduce a variety of interactive services, further strengthening its role internationally.
Another project established by CableLabs that is pertinent to the future of interactive services is VOD Metadata (www.cablelabs.com/projects/metadata). Its purpose is to examine the distribution of content assets, such as movies, coming from multiple content providers and being sent over various networks to the cable operators.
Metadata refers to descriptive data associated with a content asset package. It can be something simple such as identifying the content title, or it can be much more complex, such as providing a complete index of different scenes in a movie. The initial efforts of VOD Metadata are focused on creating specifications for VOD and subscription VOD (SVOD) applications.
CableLabs has a home networking initiative called CableHome (www.cablelabs.com/projects/cablehome). CableHome's objective is to deliver to subscribers high-quality, managed, value-added broadband services over any home network media. Furthermore, the objective is that by complying with CableHome, different manufacturers can develop interoperable products, making it as convenient as possible for the consumer and reducing time to market, complexity, and costs.
Cable operators and telcos are taking advantage of revenue opportunities provided by home networking services. Major cable operators are conducting trials or have rolled out services that enable and support providing broadband to multiple PCs. Consumers see value in these services and, as a result, cable operators are realizing incremental revenue now. In addition, forward-looking services, such as medical monitoring, energy management, and networked personal video recording (PVR) distribution, will be increasingly visible. (HANs and the specifics of the CableHome specification are discussed in more detail at the end of this chapter.)
The Future of Cable TV Networks
The industry perspective is that within 10 to 15 years, cable networks will have evolved to a converged platform. Analog video will have ceased, and telephone will have migrated to VoIP, leaving only QAM-modulated downstream channels carrying MPEG transport stream packets. Some of those packets will carry video, and others will carry IP traffic, which might include voice, computer files, video, set-top box commands, or network management traffic. All upstream communications will be based on DOCSIS, including cable modem traffic, IP voice, and communications from DOCSIS-enabled set-top boxes and gateways and system-monitoring devices. Encryption modules will be portable, allowing any set-top box to decrypt any transmission if properly authorized, and headend encryption will no longer be inseparable from multiplexing and modulation.
VOD is cable TV's fastest-growing new service category. Aside from providing additional service revenue, VOD is an essential tool in cable's competition with satellite. It exploits three inherent advantages that HFC networks have over direct broadcast satellite networks: greater raw information capacity, a broadband two-way connection, and a customer base that can be segmented into small groups. This combination allows the network to dynamically create and manage bidirectional information sessions with individual customers or groups of customers, which enables not only VOD but telephony, Internet access, and multiplayer interactive gaming.
The initial successes of VOD launches coupled with new flavors of on-demand, such as SVOD and free-on-demand (FOD), have strained network resources and raised new issues such as how to manage very large video libraries, update commercials and advertisement-supported content, and address network scaling and redundancy protection.
The eventual cessation of over-the-air and cable-carried analog broadcast television will greatly relieve the bandwidth crunch, even if some of that bandwidth is reassigned to expanding the upstream spectrum. Current estimates, however, are that this is not likely to happen until 2009 at the earliest. We have limited options for solving the immediate bandwidth problem: increasing the downstream bandwidth, decreasing the node sizes or subdividing the nodes, using more efficient digital video encoding, more efficiently sharing bandwidth between applications, using more aggressive modulation formats (such as 1024-QAM), or moving to switched real-time broadcasting. Also, advanced codecs such as MPEG-4 and Windows Media 9 (which are discussed in Chapter 10) promise a leaner bandwidth profile with potentially better picture quality. But introducing them onto MPEG-2-dominated networks isn't going to be easy or quick.
The proposed next-generation network architecture calls for a headend with the ability to transcode video, converting content there as opposed to converting it at a video storage hub further out in the network.