DSL Primer

The first practical application of high-speed data over copper wire was demonstrated in the late 1980s at the labs of Bellcore. This first harnessing of higher frequencies was offered as a one-way traffic flow, called asymmetrical. These first efforts led to the integrated services digital network (ISDN), which is historical proof that the idea of integrated voice and data is not new with DSL. As another example, data over voice (DOV) predates ISDN. Not too long afterward, high-bit-rate DSL (HDSL) was created, which is now used for most traditional tier-1 (T1) dedicated circuits in the U.S. market. 

Discrete multitone (DMT) was the next major development that assisted the creation of DSL. By separating the high-frequency signal into 256 subchannels, the problems created by line noise and other types of interference were minimized. This development opened the door for more market-ready DSL development. 

Two groups were originally formed to further the development of DSL. The first, the Asymmetric Digital Subscriber Line (ADSL) Forum, is a group of vendors and carriers dedicated to the continued development of that version of DSL. The second, the Universal ADSL Working Group (UAWG), proffered the G.lite standard (or ADSL lite), which does not require a splitter. G.lite was squarely aimed at the reduction in installation costs of implementing DSL out in the field. 

xDSL and DSL are quickly merging to become synonyms, describing the various types of DSL that exist in the marketplace. DSL enjoys a large number of enviable characteristics. For instance, DSL is a very secure service. Just like a traditional phone call, DSL is a direct link between the customer and the carrier. DSL service is referred to as always on, meaning that when asked it delivers; there are neither links to establish nor numbers to dial and answer. DSL’s speed is enviable—up to 35 times faster than a standard dial-up connection. Last but not least, DSL is able to transport voice and data services over the same copper wire at the same time. 

DSL, however, does have some drawbacks. As with any new service, the process for ordering, provisioning, installing, and testing DSL is still being developed within the industry. As mentioned earlier, there are some conditions that must be met in order for DSL to be effective (and therefore available) from a particular site. Depending on the version discussed, DSL has distance limitations. The site must be within so many feet of the serving central office (CO). ADSL is available up to 12,000 feet, symmetrical digital subscriber line (SDSL) up to 18,000 feet, and an ISDN–like version of DSL (IDSL) can reach out to 30,000 feet. What are the different versions of DSL? Table 1 provides general information on these variations in the DSL family. 

Table 1. Digital Subscriber Line Service Types

Interworkings of DSL

DSL is able to provide these increases in available bandwidth by using more of a frequency spectrum that can be transmitted across a copper telephone line. The traditional phone call travels in the 300-MHz to 3,200-Hz range. DSL uses this spectrum and more, employing digital encoding methods to maximize the amount of information that can be transmitted per unit of time. These transmissions can be used to transport voice and data services at the same time. There are restrictions, however. DSL technologies (DSL modems at the customer site, DSL access multiplexer [DSLAM] at the termination point of the service) must be installed across the entire service span, and that span (length of transmission cable) must be able to support these higher frequencies consistently. Items such as load coils or other objects that change the electrical characteristics of the line must be removed so that the line can support DSL services. 

POTS Splitter

At times and with certain services, a plain old telephone service (POTS) splitter is used (see Figure 1).

At other times, however, a splitterless installation is employed (see Figure 2). 
Figure 2. Splitterless Installation

This splitter uses a low-pass filter (which means that the filter only allows low-band frequencies to get through) to separate the voice or low-end frequencies from the data or middle- and high-end frequencies. If a splitter is not required (and some DSL services do not require such a device), then these DSL services are referred to as splitterless DSL. A DSL service that does not require a POTS splitter is easier and cheaper to provision in the network, for a carrier employee is not required to make a trip to the customer’s premises to install such a device. It is preferable for splitter designs to be passive (meaning no power is needed); in the event of a power failure, the basic phone service is still available during an emergency. 

Line Coding and DMT

DSL services can be provided with one of three line-coding methods: DMT, quadrate amplitude modulation (QAM), and carrierless amplitude/phase modulation (CAP). DMT is typically used for DSL services. 

DMT is a method of coding information (line coding) to be transmitted over the physical phone line. The information is split into a number of channels, each having the same bandwidth requirements but packaged to be transmitted at different frequencies. This method has a number of advantages, such as channel independence. If a line gets noisy at one frequency range, others may still get through; thus, the throughput is only reduced rather than stopped. 

The ANSI specification for ADSL uses 256 frequency channels for downstream transmission and 32 channels for upstream. All of these channels have a bandwidth of 4.3125 kHz, with the same number of kilohertz separating each channel from the other.

With All of This Variety, How Big Is the Market? 

A number of consulting and research houses monitor the DSL market. Various forecasts can be purchased that suggest the amount of equipment to be sold, revenues carriers might expect from services provided, and so on. To say that the DSL market is difficult to forecast would be an understatement. Several firms were quoted in January 1999 with their own forecasts as to the number of lines of DSL that would be installed by the end of that same year. All of these forecasts were exceeded by April 1999, and the growth curve has not stopped outstripping the updated forecasts. Equipment sales are expected to reach US$2 to 3 billion by the year 2005, and service forecasts are expected to reach US$2 to 3 billion by the year 2003. Actual line deployments by the summer of 1999 were pegged at nearly 200,000. With about 180 million working copper loops in the United States alone, the market size for DSL services is very large indeed. 

Once a basic understanding of DSL technologies and how they work is established, a specific kind of DSL may be discussed: ADSL. ADSL is the most common DSL variety used to support VoDSL service.

About VoDSL

VoDSL is means of leveraging the copper infrastructure to provide both quality voice services as well as support a wide variety of data applications over the same existing line to the customer’s site. It gives data competitive local exchange carriers (CLECs) a way to increase revenue potential, incumbent local exchange carriers (ILECs) an answer to the cable modem, and interexchange carriers (IXCs) a way to gain access to the local voice loop. In short, any carrier type can increase the value of its marketed services basket. 

With all the bandwidth enabled through an ADSL circuit, there are some maximums of service types that can be provided. One such example is to ask how many voice lines could an ADSL circuit provide. Depending on the type and amount of compression and the actual line speed, the answer might be surprising. See Table 2 for some example calculations. 

Table 2. DSL Dedicated to VoDSL 

This is just one example why types of carriers are excited about the possibilities offered by VoDSL. In crowded markets, where the availability of cooper loops is at a premium, the application of this kind of technology can be a shot in the arm to the carrier’s profitability. 
VoDSL also generates high revenues. A number of research houses have stated that the service revenue expectation for data-based DSL services could be in the neighborhood of US$3 billion a year, while the anticipated services revenue for VoDSL is expected to be near US$13 billion.

How Does VoDSL Work?

VoDSL requires a platform of DSL equipment, coupled with platform adaptations or additional equipment that can handle the requirements for voice services. 

Figure 3. VoDSL Service Equipment

The following are necessary to provide VoDSL: 

• customer equipment—telephones, private branch exchange (PBX), key system, fax, modem, and so forth 
• integrated access device (IAD)—The IAD can serve multiple functions, including those of a DSL modem. The IAD serves as the interface between the DSL network service and the customer’s voice and data equipment. The packetization of voice traffic takes place on this unit utilizing standards-based technology (usually asynchronous transfer mode [ATM]). The customer premises equipment (CPE) prioritizes the voice packets over data calls to ensure toll-quality voice delivery and then sends the packets over the DSL line. 
• DSL line—transports the data and packetized voice to the nearest carrier facility utilizing existing twisted-pair copper loops. Of course, these loops must be able to support the distance and quality requirements for DSL service to be offered. 
• DSLAM—terminates multiple DSL lines and aggregates traffic from them 
• data switch—receives the traffic from the DSLAM and separates the data from the voice packets; data is sent to a data network (e.g., Internet), while voice packets are sent to the voice gateway 
• voice gateway—Voice packets are depacketized and converted to a standards-based format (GR–303, TR–08, or V5.X) for delivery to a Class-5 voice switch. 
• Class-5 voice switch—telephony switch providing dial tone, call routing, and services; also generates records used for billing 
• public switched telephone network (PSTN)—the public telephone network

VoDSL broadband access solutions are now hitting the market. These types of solutions are generally overlays onto the DSL network, enabling the provisioning of voice services on what started out as a data network. 

Today, a VoDSL solution has two basic components: a voice gateway and an IAD. The voice gateway allows the traffic to be peeled off of the data network and handed to the PSTN for service and switching. The IAD provides the interface between the DSL network service and the customer’s network equipment. Note that an IAD can be used to connect both voice- and data-literate equipment. 

From a layering approach, VoDSL has several layers as well. The physical layer is actually a twisted pair of copper wires. Another is the transport layer, which can be handled by frame relay, ATM, or Internet protocol (IP). Frame-based VoDSL actually looks like voice over frame relay (VoFR); ATM–based VoDSL behaves like voice and media over ATM (VMoA); and IP transport for VoDSL is actually voice over IP (VoIP). Yet another layer is voice coding, followed by the signaling layer. The ADSL Forum is working to the idea that its standards work will concentrate on using channel associated signaling (CAS) in its first release of VoDSL work. Finally, the services layer is supported, offering services from dial tone to call waiting. The voice-capable switch in the carrier’s network supports the services layer.

Transport Methods within VoDSL 

The VoDSL market was struggling in its infancy to decide just how to transport the service. Should IP, frame relay, or ATM be used? Each method has its merits; however, all methods are moving packetized voice across packet-based networks to its ultimate destination. 

IP has ubiquity going for it; frame relay has ease of implementation (if the customer site is already a frame customer); and ATM has the strongest quality-of-service (QoS) history. Much discussion has been held within the ADSL Forum to determine which course to take. The vote was overwhelmingly in favor of using ATM technologies as the basis for the first phase standard approach to VoDSL service. For those wanting to follow this standard work, refer to WT–43 from the ADSL Forum for more information. 

Analysts estimate that about 90 percent of the installed DSLAM uses the ATM transport method back to the switch. Older techniques use ATM adaptation layer (AAL1), employing byte-interleaved multiplexing, which is sometimes called time-division multiplex (TDM) over ATM. The ADSL Forum is anticipated to adopt the more efficient AAL2 (with permanent virtual circuits [PVCs]), which uses packet-interleaved multiplexing. AAL2 is more efficient because it allows the network to allocate bandwidth dynamically on the DSL service between the demands of voice and data services. If no voice services are in use, then the entirety of the bandwidth can be dedicated to data services (unlike AAL1 with PVCs, which keeps the bandwidth needed for voice open in case it is needed). Use of AAL2 also allows for something called silence suppression, which can recover up to 50 percent of the bandwidth allocated for the voice traffic. Silence suppression removes the necessity of packetizing the silence in a phone conversation (where no one is talking) and instead inserts data into the packet stream. Table 3 provides information on where more detailed standards information may be found.

Table 3. Voice over AAL2 Standards 

VoDSL can also be moved via IP, which is a method described as voice over multiservice data network (VoMSDN) by the ADSL Forum. 

A MSDN can be exclusively IP–based or incorporate other packet technologies as well. A pure IP network would transport all traffic over IP, using VoIP technologies such as real-time protocol (RTP) (e.g., RFP 1889). A mixed IP and ATM network could use IP for data traffic and ATM for voice. In a VoMSDN example, all call-control protocols are provided outside of a traditional Class-5 switch. Interactions with the PSTN, as needed, are provided via a gateway that typically supports GR–303 to interconnect. Call control in this type of environment is performed by protocols such as H.323, H.248, or session initiation protocol (SIP).

The Future of VoDSL 

There is no doubt that the market, both business and residential, is clamoring for more and faster ways to connect to the network in this Internet age. The demand for services with DSL characteristics is well documented. The ability for the carriers serving the public networking needs to fulfill this demand must certainly be tempered by the development of standards, solutions to line-quality problems, and the always-requisite interoperability standards, to name but a few. However, there is too much demand and too many capital dollars chasing solutions to believe that the DSL installation and maintenance conundrum will not be solved in the next few years. 

One of the major drivers for those solutions is not just data, but traditional and new ways to offer voice services. Voice traffic pays the bills and provides the majority of the net income in the communications market. In short, the demand for high-speed services, coupled with the ability for voice to fund the development of future network architectures, paints VoDSL’s future a rosy color indeed.

The beauty of DSL technologies lies in its ability to enhance the capabilities of the existing copper lines to be able to transport voice and data services simultaneously, at speeds that rival dedicated services in some instances. As the numbers of VoDSL lines grow, the network will increasingly look for solutions to handle this volume of integrated voice and data traffic. 

Next-generation switching products must be able to handle this integrated voice and data traffic at the first point of switching. Additionally, more standards work must be approved to further enhance the functionality of VoDSL services. 

For instance, the ADSL Forum may very well consider more signaling requirements above and beyond the existing CAS recommendation. H.248 signaling for the access side of VoDSL is a good candidate. On the trunk side, requirements for VoDSL network switching equipment to support bearer-independent call control (BICC) may become necessary. 

Adaptation may very well see additional standards recommendations as well. For instance, any packetized voice service must contend with echo cancellation. The delay that is introduced from the packetization of voice information results in audible echo to the parties involved in the call. It is true that network design can overbuild to compensate for some of the introduced echo, but to be able to provide truly consistent VoDSL toll quality, some sort of echo-cancellation capability standard may need to be recommended for next-generation switches.