DSL Advances

Basic VDSL architecture presumes a shorter twisted-pair transmission line than earlier DSLs. VDSL line lengths are typically between 150 meters (500 feet) and 2000 meters (6600 feet). On such short phone lines, the very high-speed digital transmission of tens of megabits/second is possible. The existence of shorter phone lines implies a telecommunications network that increasingly uses fiber, or perhaps also wireless transmission, in that the nearest central-office portion of the access network previously carried all data via twisted-pair. All DSL technologies prior to VDSL had served the large majority of customers via copper wire pairs directly from the CO to the customer location. VDSL's short line reach, in contrast, requires the large majority of customers to be served via copper wire pairs running from the customer site to a nearby network node that is then linked to the central office via fiber or radio. Thus, this architecture could be described as hybrid fiber-copper .

Section 7.1.1 describes the basic VDSL combination of twisted pair with fiber and relates speed/length trade-offs for VDSL, whereas Section 7.1.2 then further details consequent architecture for VDSL. An important area of VDSL performance and installation is spectrum compatibility with an array of existing DSL and home phone “line signals, on both twisted pair and in the wireless ether surrounding telephone lines. Section 7.1.3 then illustrates some of the complexities of spectral design and describes and evaluates current VDSL spectrum recommendations. A brief synopsis of what we here call "the grand debate" is provided in Section 7.1.4 and then is validated by the two transmission methods specified in the draft standards to be described later in Sections 7.3 (DMT) and 7.4 (QAM).

7.1.1 ADSL Extension

ADSL is now acknowledged as a successful telecommunications service with tens of millions of lines in deployment, and hundreds of millions hoped to be deployed in the next decade or two. However, in its earliest days of standardization, ADSL faced the severe criticism that even its greatest standardized speed of 8 Mbps was too slow to match the data rates possible on what are called hybrid fiber-coax (HFC) networks. HFC networks upgrade existing unidirectional cable TV networks in two ways:

1. The cable TV networks are made two-way in HFC by replacing upstream-blocking unidirectional amplifiers in the cable plant by more sophisticated two-way non-upstream-blocking bidirectional amplifiers .

2. The cable TV networks are increased in bandwidth in HFC by replacing coax near the TV head-end by fiber, multiplexing multiple separate coax signals on the same fiber in different bands, and thus rendering fewer subscribers per coaxial branch.

Phone companies believed in 1994 and 1995 that they must replace their existing phone-line networks with HFC, and several attempted to do so, only to find later that the costs were prohibitive.

ADSL was already bi-directional , ^[2] but with limited speeds downstream and even lower speeds upstream. In 1994 and 1995, ADSL was perceived as unable to support the full set of video, voice, and data services necessary to compete with HFC. VDSL was proposed by ADSL proponents as a next higher-data-rate step for ADSL: If fiber can be installed in HFC, then why not install it in existing networks when there are customers ready to pay for higher speeds than ADSL, and instead use fiber-based-loop-shortening to increase the speed and symmetry of ADSL? The initial VDSL architecture of Figure 7.1 ensued as the future of DSL deployment when (and if) customers were willing to pay for more and more fiber. VDSL is an incremental alternative that leverages existing phone lines in contrast to network replacement mandated by HFC. In 1995 and beyond, VDSL's incremental deployment won increasing favor with telephone service providers and is the actual mode of choice today. Cable suppliers continue to upgrade their TV networks to HFC at significant cost, but it becomes increasingly clear that the merits of DSL will prevail for nearly all services other than (unidirectional) analog and newer digital television delivery, for which the plant architecture of cable seems still to be well suited. ^[3]

^[2] The asymmetry in ADSL allowed a longer line length for reliable transmission of a given data rate [6].

^[3] It is conceivable that Internet-based approaches to TV may allow an opportunity for DSL.

The optional splitter of ADSL is preserved in VDSL so that network- powered analog voice service can be delivered normally on the same line as VDSL. The cost of the fiber section is high, but can be divided by the number of customers served. As fiber penetrates closer and closer to the customer, that cost is shared by a smaller number of customers. Thus an important trade-off in VDSL is the length of the fiber versus the length of the remaining copper. There is no single good answer to this trade-off as it depends on applications, customer willingness to pay, transmission method, and of course cost of the fiber ”however, VDSL allows a wide range of trade-offs, as this chapter illustrates.

Figures 7.2(a) and 7.2(b) illustrate data rates for both upstream and downstream VDSL transmission on 24-gauge twisted pair (.5 mm European) versus loop length for the United States, Japan, the United Kingdom, and some other countries (998 curve) and some European countries (997 curve) DMT VDSL draft standards. ^[4] Clearly the data rates are quite high on short loops , ensuring a large individual bandwidth per user (often higher than cable networks, which customers must share due to the cable plant architecture). Although still evident, the premium paid in range loss for symmetry of data rate is less as loop lengths get shorter, and then VDSL also offers a way to offer increasingly symmetric individual service to customers. As the number of small businesses worldwide explodes, most often in urban areas where line lengths are short, the potential for symmetric support of the voice, conferencing, peer-to-peer gaming or working, "home" Web server upstream bandwidths is then evident with VDSL. Today, an increasing number of service providers consider exploring early VDSL deployment for business services, particularly, EFM support (see Section 7.5). Korea Telecom has committed to deploy DMT VDSL service starting in 2003 and is the first telco to offer commercial VDSL service that is not a trial. Five percent of Korean DSL is mandated to be DMT VDSL by 2005 (and Korea has more DSL than any country in 2002 “2003). In 2002, only a tiny fraction of the nearly 10 million businesses in the United States were connected by fiber (and a yet smaller fraction in other countries) [7]. Telephone companies are installing more fiber to businesses every year. With the advent of ATM passive optical network (APON) access, the deployment of fiber is expected to accelerate. One large fiber installer and service provider estimates they will increase this number by 2,000 business in the next two years at a cost of $1 billion [7]. In 2003, construction of residential fiber-to-the-home (FTTH) began on a small scale. Thus, VDSL will play a major role in the future service offerings to small and many large businesses before fiber connection is financially viable or completed. Other service providers still believe that support of video and television may also be viable in the future, although the economics of this application may be harder to justify versus cable.

^[4] Achievable data rates for the other "single-carrier" standard will be less, with these curves of Figures 7.2(a) and (b) as an upper bound. (See [6].)

The wealth of ADSL installations also mandates another practical requirement that VDSL service must be compatible in many respects with existing ADSL. An existing customer with an ADSL modem on his premises (perhaps in his/her portable computer) may move or travel into another area, or may live in an area where VDSL arrives, and will still want his/her ADSL modem to function as it always has. Thus, the ONU-side modem in Figure 7.1, often called the LT (line termination) in VDSL would need to support ADSL service, but would of course also allow higher speed service if/when that customer decides to purchase a higher-speed VDSL modem and the higher-speed VDSL service. Also, a customer who buys a VDSL modem will certainly want that modem to work with an existing ADSL connection at lower speed if that is all that is available. In addition to interoperation with ADSL modems, VDSL modems must also be compatible spectrally with ADSL modems that may share the same binder and with existing home-premises networks, as well as with perhaps a plethora of other standard or nonstandard systems that exist in/near the cable (for instance, HDSL, SHDSL, or nearby HAM radio). Section 7.1.3 deals more directly with this spectrum issue, whereas Chapter 8 explores the area of regulation and various newer forms of spectrum management that VDSL may exploit appear in Chapters 11 and 12.

Overall, VDSL offers a mechanism for service providers to upgrade their networks incrementally and with continued profitability to include increasing amounts of fiber, approaching an ultimate goal of a network based entirely of fiber. Telephone-line service providers have a very powerful story and future with VDSL, now following ADSL.

7.1.2 VDSL Architecture

With the VDSL incremental growth opportunity for DSL for the next century, the questions are "where, when, and how" to insert the fiber into the network. Clearly the installation of fiber should usually start closer to the service provider where the cost of a fiber can be shared over the many customers served by that fiber. Further from the central office, the fiber would service fewer customers and thus appear more costly to install per customer. Indeed today, 15 percent of the American network has what is called fiber feeder as shown in Figure 7.1. Initially, with the fiber loop carrier system being deployed today, usually only the POTS/voice connections in Figure 7.1 exist. However, VDSL or any DSL is added by placing the DSLAM at the end of the fiber, sometimes known as an optical network unit (ONU). Many phone companies have massive expenditures and fiber deployments underway to augment their ADSL service deployments so that line lengths are shortened and ensure higher ADSL speeds more reliably. It is also possible to see an ADSL DSLAM also at the ONU, which is sometimes also called a remote terminal (RT) or LT, depending on the country and the exact type of DSL deployed (but unfortunately the use of these terms is inconsistent). We use the term LT in the ensuing part of this chapter. The trend toward more fiber and shorter twisted pairs will continue with time and VDSL. Splitter circuits can be used at both ends to protect and preserve the existing analog POTS service. Additionally, the high speeds of VDSL allow multiple digital voice signals to also be carried to the customer. Within the customer premises (home or business, home is shown), a gateway is used to demultiplex the various VDSL signals and route them to the appropriate applications device, which could be a phone, computer, or television/entertainment system.

Within the central office, another demultiplexer /multiplexer can be used to extract application signals and route them appropriately. Figure 7.1 presumes a heavy use of Internet delivery, as well as Ethernet distribution within the home, but other mechanisms for such multiplexing are possible and discussed. In particular, wireless local area networks (LANs) [8] and/or home-phone distribution systems [9] have been considered .

Telephone lines often are designed to have a few intermediate points to which fiber or connection of any device is simpler than any other position along the line. Since the early 1970s, phone companies have averted deploying phone lines of length greater than 2.4 miles, a length often called a "carrier service area" or CSA. Early versions of such deployment often used what is called digital loop carrier (DLC) or subscriber loop carrier (SLC). SLCs were originally served with T1 links, and later HDSL, where fiber is today. ^[5] DLCs or SLCs multiplex several voice signals onto a single twisted pair to the RT before digital signals reconvert to analog and traverse up to the last 2.4 miles as analog POTS. Such RT points are strained in terms of the physical space available for new equipment, thermal temperatures allowed (often having no fans or air conditioning, unlike a telephone company central office[CO]), and available power to energize DSL modems. Thus, as ADSL migrates into VDSL, DSL modems need to be smaller and consume less power per line, constraints that have been facilitated by continued advances in VLSI technology since VDSL's inception. Wider bandwidths also increase radiation levels from twisted pair, mandating lower power spectral densities at higher frequencies than used in earlier DSLs. The CSA interface point typically allows speeds of up to 6 Mbps/640 kbps in either ADSL or VDSL to be deployed to all customers, a factor of four (or more) increase with respect to most ADSL rollouts of 2001, which target phone line lengths of up to four miles (the latter thus covering theoretically about 90 percent of the existing network).

^[5] One might argue "where fiber is supposed to be today," as many DLCs still use T1s or HDSLs in the digital segment where fiber may eventually be placed. This process of fiber installation to the DLC is actually occurring slower than phone companies projected , and thus copper reuse, perhaps in some coordinated overall fashion to the DLC from the central office, is more prevalent than most telcos care to admit.

Another point of service potentially is the so-called "distribution point" (or sometimes carrier-service interface [CSI]), which typically is within 3000 feet of the customer. This point is where larger cables are terminated and smaller cables servicing up to a few hundred customers begin. Usually the box at the CSI basically serves as a cross-connection point for twisted pairs. However, the entire distribution-point box can be replaced if fiber feeds this CSI point. VDSL modems placed in such an enclosure then energize the subscriber-side twisted pairs that emanate. Power and size constraints are at least as difficult at this point as at the CSA remote terminal, usually with only a small area (few square inches) and about 1 watt of available power per DSL customer.

Another intermediate point yet closer to individual customers is often called the "cabinet" or "pedestal." Usually only four to sixteen customers are served from the pedestal with individual twisted pairs emanating to these customers. The pedestal again is normally a cross-connection point for telephone lines, but fiber can be deployed to this physically accessible point, and a VDSL modem deployed there. Very high speeds are possible on the resulting phone lines of 100 meters or less, potentially 100s of Mbps or more, higher yet than current VDSL (see Section 7.5).

Placing fiber to each successive point is increasingly costly because the cost of the fiber per subscriber necessarily increases as the number of customers decreases. Considerable cable-trenching, physical labor, and placement of many ONUs may be necessary as the fiber deployment extends closer to the customer. However, in the future if the customer demands higher bandwidth, then potentially higher revenue is possible also to pay for the fiber-deployment costs. Ultimately, fiber might be run to the home or even into the home to the desk/TV-top. The key to VDSL is the incremental deployment if and where customers are willing to pay for more fiber. The cost of deploying fiber can be from $100,000 to $1 million per half mile in areas of reasonable customer density. The placement of an ONU can cost an additional $250,000. Fiber to the premises ultimately eliminates active electronics intermediate to customer and telco, a potential maintenance advantage that should be figured also.

Realistically, though some service providers hesitate in admission, fiber-fed terminals will see mixture of lines from 12,000 feet down to 100 feet, and VDSL and ADSL modems will be operating from the same remote terminal (in practice, even if not in theory).

HFT Concept

Alternatively, a concept that is analogous to HFC networks has often been promoted for VDSL. In hybrid-fiber twisted pair or hybrid fiber-coax, the fiber carries fully modulated analog DSL signals from the CO on different wavelengths , which are then demodulated and optical-to-electrical converted for final transmission on the phone line at the fiber/twisted-pair interface. This particular configuration could have considerably less power consumption and size required at the fiber/tp interface than the usual VDSL configuration in Figure 7.3. However, it may be wasteful of optical bandwidth ”with the state-of-the-art wavelength-division-multiplexing technology today perhaps limiting the number of wavelengths on the fiber (with sufficient linearity for DSL) to less than 100. However, most of the digital complexity (signal processing and multiplexing at various levels) is then remoted to the CO where it can be more easily and cost-effectively implemented. This technique remains a research subject presently.

Unbundling Issue

Colocation of VDSL modems is yet more difficult when the VDSL modems are not in a CO. This is because sharing of space by different service providers at the cabinet, CSA, or distribution point is physically difficult (there is not enough space). Today this is a hotly debated issue in DSL deployment, and a single solution has not yet emanated. Some service providers accuse incumbent local exchange carriers of installing more fiber just to complicate collocation. Potential solutions for VDSL collocation are to:

1. Standardize the backplane interface and card size(s) of VDSL so that many service providers may plug into an ONU. (This is still problematic in that someone must go to the ONU and do it, which is costly and difficult in buried ONUs)

2. Provide separate fibers to the ONU for each service provider and divide the existing small space according to the fibers that enter.

3. Use the HFT concept and collocate at the central office where more space is available.

4. Use higher-level unbundling at layer two or three in the protocol stack (see Chapters 9 and 11).

5. A universal line card that can be sold/leased by any service provider.

Other solutions may evolve . VDSL standards to date have only encompassed collocation by mandating that a single-spectrum type shall be used in all VDSL transmission types to minimize crosstalk between VDSLs. Largely, current VDSL standards are just beginning to address the intricacies of the VDSL collocation issue.

POTS Splitters in VDSL

Splitter circuits for ADSL and VDSL are described in basic detail and design in [6], Chapter 3. For VDSL, the necessity of a splitter continues to receive attention. The rising use of splitterless ADSL suggests that perhaps splitterless VDSL is also advisable for compatibility and volume deployment reasons. The first splitterless VDSL proposals appeared in [11], [12]. Although these proposals in standards met with minority opposition (which is sufficient to block standardization), most advocates of the design in Section 7.3 are pursuing various forms of splitterless operation as an additional feature and option, albeit proprietary. The VDSL technology in Section 7.4 will not operate without splitters because of the consequent home-wiring effects.

In addition to simple separation of POTS and DSL signals, splitter circuits separate internal customer wiring from DSL, typically running new CAT five twisted-pair wiring on the DSL splitter port to the DSL customer modem (without bridged taps or other internal wiring issues), while the POTS signals remain on the existing wiring. This splitter can considerably simplify transmission problems because many customers have flat (untwisted) wiring, multiple taps, and other internal wiring deficiencies that degrade VDSL transmission without splitters. At the higher transmission rates of VDSL, all these internal effects become increasingly important. A splitterless DMT VDSL modem operates in the presence of these internal-wiring VDSL effects, albeit with degradation. Nonetheless, many internal networks (particularly those in service for small businesses, but also many residences) have good quality internal wiring. Furthermore, POTS signals may be carried far more economically via digital encapsulation in the VDSL service itself. Literally, tens to hundreds of voice signals may augment high-speed data service for a customer with several telecommunications users. With such POTS delivered digitally by DSL, there is little need or use for an analog POTS line. In some high-end residences, if there is a power-failure with the internal wiring, analog POTS may still (for a single user) be returned to the phone lines automatically as long as the ONU remains capable of supporting (as always) a single line “powered analog POTS service on that same line. The potential for improved data rates, as well as easier installation, without splitters is the attraction of such operation. Such voice-over DSL has become increasingly popular, and VDSL is a logical extension of this popular nonanalog POTS DSL service. In this splitterless configuration, the VDSL customer modem becomes a gateway to both data and voice services within the premises, which may then have additional wiring for distribution.

VDSL transmission designs on a splitterless channel will need to be robust to bridged taps, increasing amounts of radio interference, potential crosstalk (on same or other lines) from home services already present on the phone lines, and further signal attenuation. Such modems may also have need for control of power-spectral density masks also to avoid excessive emission on the customer's premises that might interfere with local HAM operators, emergency radio, or other appliances.

This area is likely to be one of controversy in the future (see Section 7.1.4) because it does distinguish technologies in Sections 7.3 and 7.4 substantially. The home architectures that emerge from ADSL (see Chapter 3) will likely be those of choice for VDSL.

7.1.3 VDSL Spectrum Issues

As the highest speed DSL yet, VDSL uses the greatest amount of spectrum. Thus, it has the greatest concerns for spectrum compatibility. The issues of crosstalk and emissions from VDSL into surrounding telephone lines and radio receivers are more important and complicated in particular. Also, the crosstalk from existing services affects VDSL spectrum design and performance. Furthermore, increasingly popular home LANs (e.g., HomePNA) on twisted pair within customer premises will also complicate issues and trade-offs, as there is both spectrum overlap on the same line with VDSL as well as crosstalk issues into other VDSL lines from the home LANs. Considerable debate occurred for the design of VDSL spectrum, and there are correspondingly three internationally approved spectrum plans (presumably one selected for any specific geographic region). Unfortunately, the competitive interests between different service providers and the competitive interests between different transmission techniques have not worked to VDSL's best spectrum advantage so far, as significant compromise has occurred, sometimes for nontechnical reasons or rationale. However, this area may be reopened as VDSL spectrum-management standardization continues and as advanced transmission enhancements alter basic parameters and issues. Fortunately the three options do provide considerable flexibility for the future, as issues are revisited by technologists, marketing persons, and national regulators. This uncertainty makes this section at this time a bit difficult to write, but this book focuses on technical issues and describes the three plans, illustrating the various trade-offs. In the long-term massive deployment of DSL, the service providers and equipment/chip vendors who best comprehend all the aspects of this area will be able to garner the best business advantages in massive DSL deployment. These spectral options appear in Figures 7.4(a), (b), and (c), and are discussed in the next subsection.

Spectral Plans

The need for a fixed spectrum plan is only necessary for compatibility of the SCM plans of Section 7.4, whereas the digital duplexing of the DMT spectrum allows arbitrary placement of band edges without spectral roll-off penalty (although a 7.8 percent cyclic prefix penalty is necessary, see Section 7.3). It is possible that the plans in Figures 7.4(a) and 7.4(b) will be replaced by those that fully consider all aspects of applications and deregulation in the future. See Chapter 11 for more on spectrum management and the future of DSL. The third international spectral plan in Figure 7.4(c) encompasses the possibility of spectrum flexibility. This option is accommodated only in the DMT VDSL standard of [17]. (It originally appeared of interest only in Sweden, but at the time of this writing appears it will be selected in several Asian VDSL deployments.)

Robustness

VDSL must be able to accommodate frequency-selective disturbances, the most well known of which are bridged taps of different lengths. In this chapter, VDSL performance in the presence of bridged taps and other frequency-selective disturbances is evaluated in terms of robustness. A robust system on a line with a bridged tap shares the unavoidable degradation equitably between the downstream and upstream directions. A robust system on a line subject to mobile radio interference adapts to cope with changes in the noise profile.

Bridged Tap Robustness

Bridged taps occur in the loop plants of all operators (including countries where the operator claims to have no bridged taps) and in particular are extremely pronounced in occurrence when splitterless designs are used. Although it is impossible to avoid the effects of bridged taps completely, it is highly desirable for VDSL performance to degrade somewhat gracefully on lines with bridged taps. For a symmetric service, graceful degradation occurs if the ratio of upstream rate to downstream rate remains close to one, even if total sum data rate up and down decreases slightly. In this symmetric case, huge rate loss in one of the directions because of bridged taps is highly undesirable. In asymmetric transmission, it is desirable to maintain the ratio of asymmetry under different bridged-tap configurations.

Figure 7.5 illustrates the adverse effects of bridged taps on transmission performance. The figure shows the transfer function (in dB) of a 4050- foot loop, with bridged taps (66, 56, 46, and 36 feet long) and without bridged taps. The bridged taps cause the transfer function to exhibit notches periodically in frequency. As the bridged taps get shorter, the notches become deeper, and they move to higher frequencies. The existence of such notches (10 “20 dB deep) can seriously harm transmission.

Below the graph, two different four-band frequency plans are drawn. Note that this specific loop has very large attenuation at frequencies above 7 MHz, so the spectrum above 7 MHz is unsuitable for data transmission. Therefore, only the lower two bands would actually be used. Each frequency plan copes differently with this kind of disturbance. If plan A were used in the presence of a bridged tap 36 to 66 feet long, then upstream transmission performance would be degraded significantly, although the downstream direction would be minimally affected. If plan B were used, then the downstream transmission performance would instead be degraded. Both plans fail to be robust.

One might argue that some other fixed four-band plan would actually show more immunity to such situations. However, the bridged-tap length is unknown; it may vary from 10 to more than 100 feet. Thus, the notches may appear in almost any frequency of the VDSL spectrum. For any four-band plan spanning the VDSL frequency range, there will always be a bridged tap with such a length that performance will be degraded in only one direction. The greatest loss for such unidirectional loss with asymmetric transmission often occurs in the upstream direction because the lower upstream bandwidth loses a greater fraction of its data-carrying capability to a notch when it occurs solely in an upstream band. However, for any desired ratio of downstream and upstream data rates, there is one optimal frequency-division duplexing scheme, which one can prove attains the maximum possible robustness. The solution is to partition the spectrum into infinitesimally small bands and alternatively assign them to upstream and downstream transmission. Then any frequency-selective disturbance (such as a bridged tap) will have an equal impact on both directions of transmission. Figure 7.6 shows such a frequency plan and illustrates why symmetric service is maintained .

The implementation of this optimal scheme may prove too complex, ^[6] so suboptimal schemes with somewhat lower but still adequate robustness may have to be used instead. By interpolating between the four-band plan and the optimal plan, we deduce that a number of bands as large as possible is highly desirable. As the number of bands increases, the data rate loss caused by a bridged tap will be distributed more evenly between the two directions of transmission. The simulations to follow demonstrate this fact.

^[6] Recent demonstrations of full "zippering" have been able to suggest that, at least in some situations, large numbers of alternating up/down bands are indeed feasible with acceptable implementation.

Simulations

The simulation results that are shown below were obtained using a popular telco simulation tool. The four different frequency plans that were evaluated are shown below (numbers refer to MHz):

998 band plan A in g.993.1

Up = (3.75 “5.2, 8.5 “12)

Down = (0.138 “3.75, 5.2 “8.5)

997 band plan B in g.993.1

Up = (3.25 “5.1 , 7.1 “12)

Down = (.138 “3.25, 5.1 “7.1)

Digital duplexing 5-band plan

Up = (0.03 “0.138, 3.08 “4.78, 10.242 “17.66)

Down = complement of up

Digital duplexing 7-band plan

Up = (0.03 “0.138, 2.5 “3.5, 4.5 “5.5, 11 “17.66)

Down = complement of up

Digital duplexing 15-band plan

Up = (0.03 “0.138, 2.1 “2.5, 2.75 “3, 3.25 “3.5, 4 “4.25, 4.5 “4.75, 5 “5.5, 10.5 “17.66)

Down = complement of up

Bandplan C in g.993.1 is generally intended to be variable in frequency cutoffs and allows for the needed flexibility in future VDSL systems. See, for instance, Chapter 10 or Section 7.5. An early option has up = (0.03 “.138, 2.5 “3.75, Fx-12) and down = (.138 to 2.5, 3.75-FX, 12 and up). This is effectively a six-band plan. Presently Bandplan C has an ambiguity as to the upstream/downstream cutoff at 2.5 as to whether it is 2.5 or 3. The author interprets this to mean that this frequency can only be chosen between 2.5 and 3 MHz.

The services evaluated were for both the noise A and noise D environment of [13]. For various band plans, the reach in meters was computed both with and without bridged taps. Table 7.1 shows the resulting data rates for a 4500-foot loop.

We immediately see that using a larger number of bands always improves upstream data rate. A four “band plan has upstream data rate annihilated with the 998 or 997 plans, a particularly concerning problem for those desiring symmetric service. It is worth noting that the reach of the extra long symmetric service is improved by more than 35 percent (300 “500 meters) when more than seven bands are used. This service represents an important market segment, and might be the first VDSL service to be deployed. A more detailed set of results appears in [14]. In particular, four lines to one customer could offer 10 Mbps symmetric "10BT" Ethernet service if seven or more bands were used. Such a service, plus downstream video, might be desired by a small business (see Section 7.5).

Mobile Radio (Ingress) Noise Robustness

Mobile radio ingress noise robustness is important also in VDSL. The area is sensitive because ingress transmissions can include those used in defense of various countries. Roughly though, individuals without security clearances in various countries can learn of two types of disturbances:

1. Several narrowband analog voice signals in a band of a few hundred kilohertz that can be anywhere between 1 and 20 MHz and can move (or hop) in their general band location

2. Direct-sequence spread spectrum signals of 300 “500 kHz bandwidth (spread voice or data) with center frequency that hops throughout the 1 to 20 MHz band

Even when frequency bands are known, non-linear/non-ideal channel/receiver elements can lead to RF in other bands through various sum and difference effects on single or multiple RF interferers.

One cannot specify in advance those bands to be annihilated by radio noise, and the joint operation of VDSL and of emergency and/or defense systems might become highly preferred, especially in emergency situations or disaster areas.

Again the solution is as in Figure 7.6 to be robust with VDSL duplexing. This section models a single data signal of width 500 kHz coupling into a phone line at a level sufficient to cause loss of use of the same band on that line. This level can vary from “80 dBm/Hz to “110 dBm/Hz, depending on the length of line and other system parameters. In this case, the duplexing plans were again evaluated. The loop simulated is again a 1.1 km .4 mm loop, and Noises A and D [15] were used in addition to the radio noise, along with 20 VDSL FEXT. The worst-case position of the radio noise actually was the same for both plans, 4 “4.5 MHz. The loss is again less for the plan with more sub-bands.

Table 7.1. Bridged-tap Robustness Results for 4500 foot 26-Gauge Loop

The seven-band plan again robustly achieves over 7 Mbps symmetrically in all cases for Noise A whereas the four-band plan is only 1.4 Mbps. The relative drop in performance for the four-band plan is also larger (even though absolute data rates are smaller because of analog duplexing. Although analog duplexing is a separate effect, it is fair to include in four-band plans because these plans were advocated by proponents of systems that must use analog duplexing, whereas the larger number of bands were advocated by proponents of systems that could implement the multiple bands without exess bandwidth. For Noise D, the data rates are 2.5 Mbps for the seven-band plan and only 575 kbps for the four-band. Again the relative as well as absolute loss is larger for four-bands because it is not robust to radio noise. A larger number of bands (more than eight) can actually increase the Noise A result to over 8 Mbps symmetric.

HomePNA

The Home Phoneline Networking Alliance (HomePNA) and in-progress standard draft G.989.1 (G.pnt) of the ITU [9] specify a use of the telephone line bandwidth between 5 and 10 MHz for internal home phoneline networking. HomePNA spectrum of course overlaps VDSL, leading to signal disruption of VDSL on the same line as well as generating large NEXT into neighboring VTU-Rs of customers other than the HomePNA user.

Despite the G.vdsl and G.pnt standards being developed within the same standards committee, the frequency bands for VDSL and G.pnt overlap. This course was justified by assuming the installation of unidirectional low-pass filters for every user of a HomePNA network (presuming that the customer takes the time to locate his network interface somewhere outside his home and then properly installs the filter), even though that filter is not necessary for his internal computers to talk to one another. The authors of this book doubt the practicality of requiring all HomePNA users to install such filters before the very first VDSL could be installed in the same distribution area.

A second, more elegant solution, which does involve complexity increase for the VTU-R, appears in [23] where G.pnt signals could be canceled from a VDSL signal when on the same line or on neighboring lines as long as there were only one or two of significant amplitude. However, the method will not work for larger numbers.

Note this spectral incompatibility concern does NOT apply to Ethernet, which is typically installed on category 5 wiring which is separate and isolated by nature and design from the telco network. Thus, even though Ethernet uses the same band, there is no actual spectral overlap on physically collocated wires.

7.1.4 The Grand Debate

Dating to the days of ISDN and ADSL standardization, there has been a debate over the best transmission technology to use for DSL. Although ADSL standards have universally selected the specific multicarrier transmission method known as DMT after considerable deliberation and testing, a few CAP and QAM proponents nevertheless marketed nonstandard ADSL modems for a significant time period, before most switched to and supported standardized DMT. Subsequent debate was heated in the marketplace, and there were abortive attempts to reverse ADSL standards (from DMT to CAP/QAM). The supposed threat of fundamentally high complexity of DMT leading to high prices eventually was unequivocally refuted in the ADSL marketplace , where low-cost interoperable components abound today. Early proposals for VDSL recommended an extension to the ADSL DMT standard (see [2]). Later, other VDSL proposals argued for standardization of the QAM/CAP technologies. A protracted and complex debate emerged. Should the VDSL standard specify DMT or SCM? Should the standard specify both techniques or only one? Should a compromise such as filtered multitone (FMT) be adopted? The chairpersons of T1E1.4, ITU-Q4/15, IEEE 802.3, and ETSI TM6 coordinated their efforts with firm determination to converge on a standard specifying one technique. Despite the best efforts of the combined leadership of all three committees , a few key companies opposed the selection of any one technique. This denied the industry a timely standard that would have fostered interoperability and would have assured both vendors and carriers that the technology they pursued was not a dead-end technology. Because a definitive standard was not possible, T1E1.4 chose to the develop a trial-use VDSL "standard" that was essentially a temporary catalog of two leading VDSL proposals.

This newer debate has continued in VDSL for many years with two large industrial consortia emerging with two complete transmission specifications for VDSL:

1. VDSL Alliance ” Discrete Multitone transmission (DMT) with digital or analog duplexing

2. VDSL Coalition ” "Single" Carrier modulation (SCM) with analog duplexing

(The single is in quotes because the VDSL Coalition actually advocates a solution with two carriers in each direction, which might possibly be augmented an optional third carrier upstream below 138 kHz.) Both groups have contributed a draft trial-use standard to T1E1.4, which appear in three documents: A common reference document [15], an SCM document [16], and a DMT document [17]. The DMT specification supports up to 4096 4.3125 kHz ADSL-style tones, and any number of up/down stream transmission bands. The lower 256 tones are exactly the same as ADSL, facilitating backward compatibility. Both groups have about fifty companies in them, with about ten common members . The companies in both groups represent an enormous cross-section of the telecommunications world.

The VDSL Alliance solution is backward interoperable with ADSL and has taken greater time to develop into its present converged state [17], but offers some outstanding flexibility and performance features in a large number of possible configurations. The VDSL Coalition specification [16] is slightly simpler to understand (although more pages in reality) and to design to, but not interoperable with ADSL. The Coalition specification had the advantage of earlier availability of transmission components that were partially compliant with it. The T1E1.4 group will revisit the trial-use standard for permanency in 2003, when it is likely only one solution will survive; thus, the grand debate will likely continue for some time.

There are a number of differences in markets addressed, data rates and capabilities, and potential chip complexities (although experts on both sides are now agreeing even at VDSL speeds, transmission modems are becoming a negligible fraction of overall DSL implementation complexity in VLSI). This area is controversial , and so the reader is better left to evaluate the two approaches themselves rather than for these authors to here opine the eventual surviving method.