March 9, 2011, 2:06 p.m.
posted by ksch
DSL is an entire family of technologies, all of which provide digital transmission over the copper wires used in the local loop, or last mile. The xDSL technologies offer many home users their first taste of broadband, and most users find that once they've tried broadband, they'll never go back. It's exciting how much broadband can improve performance, despite the fact that today's broadband is not as fast as the data rates we will see in the coming years. As discussed in detail in the following sections, the xDSL family of standards includes a large variety of speeds and distance specifications:
How DSL Works
The history of DSL goes back to 1988, when Bellcore (which is now Telcordia; www.telcordia.com) created DSL as a technique to filter out the incessant background noise on copper wires and to allow clearer connections through the use of electronic intelligence in the form of DSL modems at either end of the twisted-pair line. In essence, the engineers came up with a way to carry a digital signal over the unused frequency spectrum available on the twisted-pair running between the local exchange and the customer. By using the unused spectrum, DSL could use the basic telephone line to carry digital data without interfering with traditional voice services. Initially, the incumbent local exchange carriers (ILECs) were not thrilled with the idea; after all, it would not be as profitable as installing a second telephone line for customers seeking to access the Internet while maintaining their voice connections. In addition, offering a broadband connection stood to cannibalize the existing ISDN customers.
In the late 1990s, when cable TV companies began to offer broadband Internet access, things changed: ILECs suddenly became very interested in DSL technology, largely due to competitive pressure. Today, DSL is the main competitor to cable modem access when it comes to providing high-speed Internet access to residential users worldwide. As discussed later in this chapter and in Chapter 10, "Next-Generation Networks," service operators are very interested in providing triple-play services (combining the delivery of voice, data, and video) as well as quadruple-play services (which adds wireless services to the mix), and in order to do so, high bandwidth is mandatory, particularly in the support of HDTV.
Characteristics and Properties of DSL
A number of factors determine whether you get the kind of service that DSL promises and that you expect. The performance of any of the DSL alternatives is highly dependent on the loop length as well as the loop condition because a host of impairments can affect the performance of a given pair of copper wires. DSL modems generally range up to a maximum of about 3.5 miles (5.5 km), although new specifications are constantly being created to increase the permitted distances. The general rule of thumb with DSL is that the greater the distance, the lower the performance, and the shorter the distance, the greater the data rate possible. The original ADSL standards, using twisted-pair copper, can deliver up to 7Mbps downstream over a distance of about 1.25 miles (2 km). The latest ADSL standards, such as ADSL2+, can deliver up to 24Mbps downstream, but this rate depends on the distance between the customer and the DSLAM, the point at which numerous subscriber lines are first terminated and aggregated before connecting to the PSTN (for traditional circuit-switched voice) or packet-based backbone (for data and multimedia) local exchange and then onward to the PSTN or packet backbone; in order to deliver 24Mbps downstream, the maximum distance allowed is roughly 1 mile (1.5 km). With one version of VDSL2, VDSL230MHz Short Reach (SR-VDSL230MHz; the highest-performance DSL standard), it is possible to achieve 100Mbps in both the downstream and upstream directions, but the maximum distance is only 0.3 mile (0.5 km).
A number of factors can affect the loop condition for a subscriber, including the following:
Another characteristic of DSL is that it is a point-to-point connection that is always on. So when you have access to your ISP through a DSL line and your computer is powered up, the connection is on throughout the day. This has security implications: It is very important that you incorporate some form of firewall and security software to prevent the potential activities of a curious or malicious attacker.
DSL provides high-bandwidth transmission over copper twisted-pair. It uses efficient modulation, or line-coding, techniques that enable it to carry more bits in a single cycle (i.e., Hertz) than older twisted-pair. It uses echo cancellation, which allows full-duplex transmission to occur over a single electrical path, and it relies on frequency splitting to enable you to derive separate voice and data channels from one wire. DSL also retains power in the event of a power failure; if the electricity goes out, you lose your high-speed data services, but you retain your voice services.
Finally, it is important to consider one more factor before selecting the broadband access technology that will work best: contention ratios. In the case of DSL, contention ratio refers to the fact that users of the system compete for use of the same facility at the same timein this case, the number of people connected to an ISP who all have to share a set amount of bandwidth.
The following sections look at each of the DSL family members in turn. Figure summarizes some of the characteristics of xDSL. Keep in mind that the rates shown in Figure vary depending on the loop length, loop conditions, contention ratios, and so on.
HDSL is the oldest of the DSL technologies. It has been in full use for over a decade, and it is most commonly used by telcos to provision T-1 or E-1 services. HDSL allows carriers to do so at a reduced cost because it does not require special repeaters, loop conditioning, or pair selection in order to deliver those services. HDSL reduces the cost of provisioning T-1/E-1 because of the way it delivers the bandwidth (see Figure). A traditional T-1/E-1 environment makes use of two twisted-pairs. Each pair carries the full data rate, which is 1.5Mbps with T-1 and 2.048Mbps with E-1. Because each pair is carrying such a high data rate, higher frequencies need to be used; as a result, repeaters need to be spaced roughly every 0.5 to 1 mile (900 to 1,800 m). Furthermore, no bridge taps are allowed in the traditional T-1/E-1 environment.
1. Traditional T-1/E-1 versus HDSL provisioning
HDSL is a symmetrical service, meaning that it provides equal bandwidth in both directions. In addition, it is full-duplex, so it allows communication in both directions simultaneously. HDSL modems contain some added intelligence, in the form of inverse multiplexers. Because of these multiplexers, each pair carries only half of the data rate (and the receiver combines the two to deliver the full data rate), and as a result, those bits can ride in the lower range of frequencies, thus extending the distance over which they can flow without the need for a repeater. The allocation of bandwidth on HDSL depends on whether it is operating on T-1 or E-1 capacities; in the T-1 environment, it offers 784Kbps on each pair, and in the E-1 environment, it offers 1.168Mbps on each pair. With HDSL, you need a repeater only every 2.2 miles (3.6 km) or so. In addition, bridge taps are allowed with HDSL. These factors reduce the cost of provisioning services to customers and allow more customers who are outside the range of the traditional T-1/E-1 environment to enjoy the privileges of this high-bandwidth option. Because taps can be used with HDSL, provisioning can occur rather quickly. Also, HDSL is a good solution for increasing the number of access lines via the digital loop carrier transport because it is compatible with the existing loop carriers. Key applications of HDSL include replacement of local repeatered T-1/E-1 trunks, use as a local Frame Relay option, use in PBX interconnection, and use in general traffic aggregation.
HDSL is largely used to provision digital services to business premises. It is standardized under ITU G.991.1 and ANSI T1E1.4, Tech Report 28.
The HDSL2 (two-wire) and HDSL4 (four-wire) specifications were developed in order to provide the capacities and symmetry of HDSL to residences. HDSL2 involves the use of a single twisted copper pair for distances up to 2.2 miles (3.6 km). HDSL2 is a symmetrical, full-duplex service that offers up to 1.5Mbps in each direction. HDSL2 does not support voice telephone service on the same wire pair. The main difference between HDSL2 and HDSL is that HDSL2 uses one pair of wires, whereas HDSL uses two pairs. HDSL4 makes use of two pairs and also supports 1.5Mbps in each direction. HDSL4 is the same as HDSL2 except that by using two pairs of wires, it can achieve a 30% increase in the distance allowed.
Today, newer standards than HDSL, such as G.SHDSL (discussed later in this chapter), are preferred and used in new installations.
SDSL involves a single twisted copper pair that can be up to 3.5 miles (5.5 km) long. It is a symmetrical, full-duplex service, and it supports multiple data ratesup to T-1 or E-1 ratesso you can subscribe to varying bandwidths, up to 1.5Mbps or 2Mbps. Symmetry can be very important, depending on the application. If your only goal is to surf the Internet and browse Web sites, then most of the bandwidth you will need is in the downstream directionfrom the network to youin which case solutions such as ADSL are appropriate. But if you're telecommuting or operating in a small office/home office (SOHO) and you need to transfer large files or images or engage in videoconferencing, you need a great deal of bandwidth in the upstream direction as well as in the downstream direction, and in these cases, symmetrical services are best. So if your major purpose for wanting broadband access goes beyond Internet surfing, to a use such as increased productivity with school or professional work, then SDSL is probably a better option than ADSL. It is more costly than asymmetrical options, but it provides a better performance guarantee.
Applications of SDSL include replacement of local repeatered T-1/E-1 trunks, use as fractional T-1/E-1, interconnection of PBXs, support of multirate ISDN, support for switched 384Kbps service (and therefore appropriate bandwidth for lower-level videoconferencing), support for local Frame Relay options, traffic aggregation, and high-speed residential service.
SDSL was not standardized until the ITU-T standardized G.SHDSL (discussed in the following section). Unfortunately, this leads to some confusion because in Europe G.SHDSL was standardized by the European Telecommunication Standards Institute (ETSI) under the name SDSL. This ETSI variant is compatible with the ITU-T G.SHDSL regional variant for Europe. It is important to be aware that equipment referred to as supporting SDSL is generally proprietary equipment that talks only to SDSL equipment from the same vendor or another vendor that uses the same DSL chipset. However, most new installations use the standardized G.SHDSL equipment instead of the older SDSL.
Two main drivers influenced the introduction of G.SHDSL. First, the industry observed that there was a need for a higher-speed digital transport service for business applications. HDSL was available but had never been adopted as an international standard. Second, SDSL, which was introduced in the late 1990s as the business-class DSL service, never became a standard, and it also posed problems in that it interfered with the residential ADSL service, being very noisy and therefore spectrally incompatible with ADSL (i.e., creating difficulties when deployed in the same cable bundle as ADSL). A global standard was therefore needed. G.SHDSL was developed to incorporate the features of other DSL technologies, such as ADSL and SDSL, and it can transport T-1, E-1, ISDN, ATM, and IP signals.
G.SHDSL, often referred to as simply SHDSL, was the first international standard for DSL, originally ratified by the ITU-T in February 2001 in recommendation G.991.2. The standard was then updated in 2003, and that version is referred to as G.991.2bis. It is compatible with the ETSI SDSL and ANSI HDSL2 standards. This is more important than it may seem on the surface: A worldwide standard allows for global mass deployment, and higher volumes worldwide mean lower equipment prices.
G.SHDSL is a symmetrical service with options to operate over one pair or two pairs of copper wires, and it also has rate-adaptive capability. Symmetric bandwidth, an increasingly desirable characteristic, supports advanced applications that require high performance in both directions. G.SHDSL supports data rates up to 5.6Mbps in each direction. G.SHDSL eliminates the need for T-1/E-1 repeaters on loops under 3.5 miles (5.5 km), but because the technology supports the use of signal repeaters, users outside the range of ADSL can now be offered DSL service. G.SHDSL also offers improved reach, providing a 20% to 30% increase compared to HDSL and SDSL. On the average, G.SHDSL provides 3,000 feet (1 km) increased reach over previous symmetric technologies such as SDSL. To be fully appreciated, this has to be put into the context of the serving area: An increased reach of that amount translates to approximately 40% increase in coverage area. Of course, a greater serving area means more customers served, which means more revenue potential for service providers.
Unlike SDSL, G.SHDSL is spectrally compatible with ADSL, causing little noise or crosstalk between cables, which means G.SHDSL can be mixed in the same cable bundles with ADSL, HDSL, HDSL2, and IDSL without much, if any, interference. This maximizes the deployment options available to service providers, allowing G.SHDSL and ADSL to be deployed from the same platform.
G.SHDSL supports a wide range of business and residential applications that demand high bit rates in both the downstream and upstream directions. The voice and data applications that businesses can benefit from include the following:
Residential applications of G.SHDSL include the following:
G.SHDSL also supports a third market, the multiunit market, generically called MxU and alternatively referring to as multiple-dwelling unit (MDU) or multiple-tenant unit (MTU). MDUs include apartment buildings, condominiums, and commercial multitenant office buildings, and MTUs are generally hotels. The G.SHDSL with Inverse Multiplexing over ATM (IMA) feature enables G.SHDSL over multiple lines multiplexed together to offer higher-speed service rates between MxU and the network without needing to install T-3/E-3 lines or pull fiber to the building.
The main drawback of G.SHDSL is that its data rates are not sufficient to support triple- or quadruple-play applications.
ADSL was initially introduced in 1993, with the principal driver being the much-anticipated deployment of video-on-demand (VOD). However, because of some early issues with video servers, including storage capacity and processing power, VOD was largely abandoned at that time. On the other hand, Internet access was emerging as a highly desirable service, and with 80% of the traffic flow on the Internet being in the downstream direction, ADSL presented a perfect solution. ADSL, also referred to as G.dmt, was ratified by the ITU-T under recommendation G.992.1 in 1999. It is also standardized under ANSI T1.413, Issue 2.
The asymmetric operation is the key distinguishing characteristic of ADSL. Traditional ADSL, operating over a bandwidth of 1.1MHz, supports downstream data rates from 256Kbps to 7Mbps and upstream rates from 64Kbps to 800Kbps. The maximum reach is 3.5 miles (5.5 km). There is also a version of ADSL known as ADSL Lite, or G.lite. By limiting the data band to 550KHz, ADSL Lite operates at lower speeds, typically up to 1.5Mbps downstream and up to 512Kbps upstream. It is rarely used today. Again, remember that the real-world performance of ADSL, like that of any other DSL, depends on a number of factors, including the distance from the CPE to the DSLAM, the signal-to-noise ratio, the signal attenuation, the cable diameter, the line impedance (which is affected by changes in weather as well as the number of taps on the cable), and the general condition of the cable.
Because ADSL carries a mixture of traffic, including voice, data, Web access, and multimedia, a multiplexing technology is required to carry both time-critical and less time-critical data. ADSL is therefore commonly deployed with ATM, which serves this purpose. In addition, because different ATM virtual circuits (VCs) can be allocated for different services, ATM ensures that service providers can provide triple-play services. However, recently, some network operators have been moving away from the use of ATM, for reasons of cost savings, and replacing it with Ethernet-based solutions. ADSL service providers offer either static or dynamic IP addressing. Static IP addresses are preferred in cases where the subscriber wants to connect to his or her enterprise via a VPN or to host a Web server.
ADSL allows for simultaneous voice and Internet traffic on the same twisted-pair that used to be a phone line. It reserves the bottom 4KHz of spectrum for the voice traffic, and filters (known as splitters) at each end of the copper pair split the frequency bands. The lower frequencies are sent to the local exchange to switch the voice traffic. The higher frequencies are sent to the DSL modems, and a user is generally connected over a packet-switched backbone to the ISP.
Carrierless Amplitude Phase Modulation (CAP) was the de facto modulation scheme for ADSL deployments, used in 90% of the installations, until 1996. However, the DMT modulation scheme was selected for the first ITU-T ADSL standards, G.992.1 (also called G.dmt) and G.992.2 (also called G.lite), and CAP is no longer used. DMT, which is an Orthogonal Frequency Division Multiplexing (OFDM) technique, is a multicarrier scheme in which the spectrum is divided into 256 4KHz carriers. Variable numbers of bits are put on each carrier, and the portions of the frequency band that suffer interference from other devices don't have any bits put onto them. The result is improved performance. Compared to CAP, DMT is less prone to interference and can carry data over a longer distance.
Figure shows an example of an ADSL environment. At the residence, a splitter is splitting off the plain old telephone service (POTS) to the telephone instrument, using the bottom 4KHz of spectrum on the twisted-pair; the remainder of the line is left for the ADSL modem and the data communications. At the top of the figure is a business environment using a DSL line to carry the voice and data traffic, on an integrated basis.
2. An ADSL configuration
In Figure, numerous DSL lines come in from residential and business premises. Their first point of termination is the DSLAM. The DSLAM is a network device, usually located at the telco local exchange or within a neighborhood digital loop carrier configured to support DSL, although it can also reside at the customer premises in the case of a large enterprise or an MDU. The service provider generally connects all its DSLAMs, which are essentially massive ATM concentrators, over an ATM backbone network. DSLAMs are designed to concentrate hundreds of DSL access lines onto ATM or IP trunks connecting to ATM switches, routers, or multiservice edge switches that then connect the DSLs to the ISPs. Each DSLAM has multiple DSLAM aggregation cards, and each card has multiple ports (typically 24, although the number can vary by manufacturer). DSLAMs typically contain power converters, DSLAM chassis, DSLAM cards, cabling, and upstream links. The upstream links are most commonly either multigigabit fiber-optic links or Gigabit Ethernet. The DSLAM splits the voice and data traffic, sending the voice traffic through traditional local exchanges onto the PSTN and sending the data traffic to the appropriate ISP or enterprise network. Each subscriber's traffic appears as a separate ATM VC (a permanent virtual circuit [PVC]). If the service provider provides access to more than one ISP, an ATM VC is created to provide the subscriber with access to his or her ISP of choice. DSLAMs also support quality of service (QoS) features such as priority queues, contention, and DiffServ.
In July 2002, the ITU-T approved G.992.3 and G.992.4, two new standards for ADSL technology collectively called ADSL2. The basic goal of both standards, known as G.dmt.bis and G.lite.bis, respectively, is to increase the transmission rates, increase the range, and improve the overall reliability and manageability of DSL services.
ADSL2 adds several new features and functions, all aimed at improving performance and interoperability, while adding support for new applications, services, and deployment scenarios. The key changes include improvements in the data rate and reach, rate adaptation, diagnostics, and standby mode.
ADSL2 supports higher data rates of up to 12Mbps downstream and 1Mbps upstream, depending on loop length and other factors. It also improves the reach by about 600 feet (180 m). ADSL2 introduces a number of features, including power cutback capability (i.e., two power modes that help reduce overall power consumption while maintaining ADSL's always-on functionality for the user); reduced framing overhead; better modulation efficiency (i.e., seamless rate adaptation [SRA], which allows modems to signal one another in the event of changing transmission conditions and adjust the data rates almost instantaneously); channelization capability (i.e., the ability to split bandwidth into different channels with different link characteristics for different applications); and bonding of lines (i.e., joining of several copper pairs to create one logically larger pipe). ADSL2 also defines an optional "all-digital mode" where the voice transmission band (25KHz) can be used to provide an additional 256Kbps of upstream transmission capacity. Other features of ADSL2 include the following:
The ITU-T approved ADSL2+, or ADSL2plus, as G.992.5 in 2003. ADSL2+ doubles the downstream frequency band to 2.2MHz, increasing the DMT channel count to 512. This effectively doubles the maximum downstream data rates, achieving up to 24Mbps on phone lines as long as 1 mile (1.5 km) while supporting up to 1Mbps upstream. However, under most practical scenarios, speeds are in the 15Mbps to 20Mbps range when delivered approximately 1 mile (1.5 km) from the local exchange.
ADSL2+ solutions are most commonly multimodal, interoperating with ADSL and ADSL2, as well as with ADSL2+ chipsets. There is also an option in which customers served by digital loop carriers could be designated to operate only on the higher frequency channels (i.e., 1.1MHz to 2.2MHz) to limit interference with customers on longer loops who would be assigned the regular ADSL/ADSL2 channels (i.e., 1.1MHz).
ADSL2+ allows service providers to evolve their networks to support advanced services, such as video, in a flexible way, with a single solution for both short-loop and long-loop applications. ADSL2+ includes all the feature and performance benefits of ADSL2 while maintaining the capability to interoperate with legacy equipment. The other major enhancement with ADSL2+ is improvement in crosstalk and interference control. The standard specifies a set of upstream and downstream power spectrum density (PSD) masks that define methods for shaping the DSL transmission signal. Particularly important on longer loops (i.e., 2.5 to 3.5 miles [4 to 5.5 km]), the PSD masks allow the modems to optimize performance by adjusting the power levels on the various DMT channels. The operator can specify which mask to use when the link is installed, or the modems can determine which mask to use in the initialization and handshaking.
Some telcos are already beginning to use ADSL2+ to support Internet Protocol TV (IPTV) services, with an eye toward HDTV. However, despite improvements in data rates and reach, ADSL2 and ADSL2+ are still not sufficient to support applications such as multiple HDTV channels; such uses require the use of VDSL, which is discussed later in this chapter. Another problem with ADSL2+ is that although the standard is approved and chipsets are plentiful, the equipment for multiple-vendor setups still requires interoperability testing.
To expand carriers' addressable markets, the ITU has developed a reach-extended version of the ADSL2 specificationcalled ADSL2-REthat allows DSL systems to reach up to 3.75 miles (6 km). This equates to more than a 20% increase in coverage and opens the door for carriers to sign up new subscribers. The ITU-T ratified ADSL2-RE under recommendation G.992.3 in 2003. While it can support up to 8Mbps downstream and 1Mbps upstream, when taking advantage of its main feature, ADSL2-RE extends a 768Kbps downstream service by approximately 0.5 mile (1 km), to 3.5 miles (5.5 km).
RADSL, a nonstandard version of ADSL, adapts data rates dynamically, based on changes in line conditions. It can therefore operate over a wide range of loop lengths and conditions, up to a maximum of 3.5 miles (5.5 km). With RADSL, the modem adjusts the upstream bandwidth to maintain a certain speed on the downstream channel. If there is a large amount of line noise or signal degradation, the upstream bandwidth may be significantly reduced, down to 64Kbps. RADSL can be configured to operate in either symmetrical or asymmetrical mode. The downstream rates range from 600Kbps to 7Mbps, and the upstream rates range from 64Kbps to 1Mbps. Most RADSL devices rely on Discrete Multitone (DMT) encoding.
The ITU-T standardized VDSL, the highest-performing DSL option, under recommendation G.993.1 in 2004. Conventional VDSL relies on a single twisted copper pair, operating over very short distances. The loop length range is just 1,000 to 5,000 feet (300 m to 1.5 km), with performance degrading over longer distances. With its greater bandwidth of 12MHz, the maximum data rates are up to 55Mbps downstream and up to 15Mbps upstream. Actual speeds vary, depending on distance and configuration. The following are some possible scenarios:
VDSL is a very high-capacity technology, and its performance degrades rapidly over distances, so it is really meant to be almost a sister technology to fiber-to-the-node (FTTN), discussed later in this chapter, going the very short distance from the curb to the home. (Or in the case of fiber-to-the-home/fiber-to-the-premises [FTTH/FTTP] with an MDU, VDSL could be run over twisted-pair from the fiber coming into the building to each apartment.) The main application for VDSL is short-loop environments, such as MDUs or MTUs.
VDSL2 standardization efforts started in January 2004. Consent and approval of the ITU-T G.993.2 recommendation were reached in 2005. The key applications for VDSL2 are the next generation of TVVOD, DTV, HDTV, and forms of interactive multimedia Internet access. (Chapter 10 provides more information on TV standards.) VDSL2 uses the same DMT modulation as ADSL, with two bandwidth options:
VDSL2 supports a larger variety of services, including integrated QoS features, the ability to carry ATM as well as Ethernet payload, and channel bonding for extended reach or rate. An additional benefit of VDSL2 is that it is compatible with ADSL, ADSL2, and ADSL2+, whereas the original VDSL (VDSL1) is not. ADSL2+ backward compatibility makes VDSL2 deployment much more attractive for service providers and will speed the adoption of this technology. Given its improved data rates and reach, power features, and QoS features, VDSL2 also enables triple- and quadruple-play applications. A typical VDSL2 connection can support at least three DTV channels, 5Mbps Web surfing, and Voice over IP (VoIP).
For all these reasons, VDSL2 is viewed as the ultimate DSL standard, being a natural evolution of ADSL2+, allowing continued exploitation of copper plants, and providing sufficient bandwidth for advanced applications. Rather than building fiber all the way to the premises, VDSL2 will give telcos the ability to support multiple standard-definition and high-definition video streams via copper (using advanced compression). Adoption of VDSL2 appears to be moving quickly, with major activity worldwide. The first chipsets and systems are already available. The major business decision will be whether it is less expensive to extend fiber or to shorten copper loops for performance gains.