DSL page

Most of us bought our computers intending to connect to the Internet, but we didn't give much consideration to the speed of our connection. We get connected at whatever speeds we can afford. For most people this means 33.6, 56kbits/s, or possibly an ISDN line. There are much faster connections available, primarily T1, but considering the expense this is not for the individual user. There are faster connections, though, that are about to become more readily available and hopefully more affordable, too.

The Telecommunications Act of 1996 has opened up the Internet Access market to different branches of the communications industry. This has led to more competition, primarily from the cable T.V. companies. In hopes of entering each others markets, the players in the communications industry have been hard at work developing ways of offering high speed information services using their existing networks. Using their existing networks, the players in the communications industry can hope to offer services in the near future, without having to perform major upgrades to the existing infrasutructure. This has led to the introduction of new possibilities for connection to the Internet.

For the telcos, the best technology they have to offer right now are digital subscriber lines, or Dsl. Telcos are hoping to offer Dsl in the near future, as just about every major telco is running a Dsl trial or they are beginning to make Dsl available.

I would like to take a look at the digital subscriber line, or Dsl, and determine just what a Dsl is. To do this, it is necessary to provide a look at the telephone network and the local loop, show some of the history behind the development of Dsl, and show some of the difficulties in implementing Dsl in the local loop. There are also several variations of Dsl, which will be discussed. Finally we will take a look at the availablitlity of Dsl and if it is actually a viable alternative to use for Internet access.

We have been experiencing a bottleneck in the local loop of the telephone network. The fiber backbone of the network can handle data rates up to 500 Mbits/s. Our analog modems can handle speeds around 33.6 Kbits/s. The problem has been that the telcos haven't been able to bridge the gap between the backbone and the home using the local loop. The telcos realized, sort of by chance, that the digital subscriber could possibly be a way to bridge the gap, thereby allowing for the use of the local loop for megabit services. This discovery has become the hope of the telcos in the Internet access market.

The telcos have put a lot of time and money into upgrading the network infrastructure. The central offices(COs) are connected by a fiber network. The telco can cost justify such an expenditure on upgrading backbone infrastructure because the costs can be spread around among all telephone subscribers. The costs of upgrading the local loop infrastructure to fiber, though, cannot be cost justified because many telephone subscribers are only interested in voice services, which can easily be provisioned over existing loops. Another problem with upgrading local loops to fiber is that there are 150 million phone subscribers in the U.S. and upgrading the local loop infrastructure will take time. However, enough subscriber demand for broadband services has led telcos to search for more immediate solutions using existing local loop infrastructure.

The solution that the telcos came up with was the digital subscriber line. Dsl is actually a local loop access technology and Dsl equipment is implemented in the local loop. In order to understand Dsl and how it works it is necessary to understand how the local loop was set up for voice services.

Bundles of cable are carried from the central office(CO) to Feeder Distribution Interfaces(FDIs) from which the local loop is extended to the individual telephone subscriber. This worked fine for telephone customer lines up to the 18,000 ft range. In many instances, however, this led to local loops that were too long and led to attenuation of signals. In order to solve the problems of attenuation, loading coils were added to boost the signal and remote terminals were installed to shorten line length.

The loading coils have been found to be detrimental to Dsl technologies. Dsls that require higher frequencies are not compatible with local loops that have loading coils installed. In order to make these lines compatible, the coils need to be removed, which is time consuming and costly.

Remote terminals, however, are conducive to Dsl and the local loop. Remote terminals were created because of the complexity of local loop infrastructure. In order to simplify the local loop wiring, the CO ran cable to the FDI and then to a remote terminal. From the remote terminal individual dedicated lines(local loops) were run to individual subscribers. Signals are collected at the remote terminal, bundled up and sent to the CO using fiber or a T1 connection. This in effect shortened local loops and made them more compatible with line length sensitive Dsl technology(as will be described later)

Another issue with making Dsl compatible with local loops is that POTS(plain old telephone service) is designed to carry voice conversation over the local loop using frequencies between 0 to 3400 hertz. This is fine for voice, but Dsl needs higher frequencies for megabit rates. This is another factor of the local loop that the telcos have addressed in implementing Dsl over the local loop.

Digital subscriber lines are actually modems. One modem is placed at the subscriber end of the telephone line and the other is placed in the telco's central office (CO). Using advanced modulation techniques, digital signals are sent back and forth on this line creating, none other than a Dsl. It is important to remember that this is implemented over the local loop, also known as the last mile.

There are several different types of Dsl coming available. The underlying technology remains the same, independent of which type of Dsl is being implemented.

The basic problem facing implementing Dsl in the local loop is getting around the 3400 hertz limitation that has been designated for voice signals. By eliminating the 3400 hertz boundary, the much broader range of frequencies can be used to transmit digital signals.

This doesn't mean that the 3400 hertz limit can be ignored, though. Problems do arise that must be dealt with when eliminating the frequency limitation. The main two being attenuation and cross talk.

Attenuation is the weakening of the signal as it passes along the copper wire. The higher the frequency at which a signal is sent the shorter the distance the signal can travel before the signal attenuates. Therefore, lower frequencies are desirable because signals can be sent over longer local loops. Higher speeds, though, require higher frequencies. This is a difficult situation for scientists who develop Advanced Modulation Techniques in an attempt to offer higher speeds at lower frequencies.
In order to combat attenuation of the signal, scientists have developed advancements in line codes and signal transmission techniques. AMI, the one of the first attempts at solving the attenuation problem in the local loop, assigned 1 bit per waveform. For example, to transmit information at 160kbits/s frequencies up to 160,000 hertz would have to be used. Higher frequencies lead to more attenuation. In order to decrease frequencies and extend loop length, scientists developed 2B1Q. This line code technique assigns 2 bits per waveform. To transmit 160kbits/s in this situation the lower 80,000 hertz would be used. This is a lower frequency than the 160,000 hertz required for AMI, therefore 2B1Q can be used to extend loop length at which information transmission can occur.

About the same time as 2B1Q was being developed, Carrierless Amplitude and Phase modulation (CAP) came on the scene. Like 2B1Q, CAP could assign multiple bits to a waveform and lower the frequency necessary to carry a signal. CAP was an advancement, however, in that from 2 to 9 bits could be assigned per waveform. This lowered the frequency, lessened attenuation and extended the length of the local loop over which CAP coded signals could be transmitted.

Cross talk is also a major concern when implementing Dsl into the local loop. Crosstalk occurs due to the structure of the telephone network. Copper cable pairs are bundled together into what are called cable binders. Within this cable binder, modulated signals being carried by individual loops can radiate energy onto adjacent loops causing distortion of the waveforms. When this cross coupling of signals occurs it is known as crosstalk.

Crosstalk is a significant problem in some situations and less of a problem in others. One form of crosstalk, know as Next or near end cross talk, is a major problem when implementing Dsl over the local loop. NEXT refers to crosstalk at the signal source where it is the strongest. Because of the strength of the signals, crosstalk is more likely to occur on a level that distorts the signal beyond repair. Far end crosstalk, or FEXT, is not as much of a problem in that signals have attenuated to a level where distortion of other loops in the same bundle is negligible. Depending on the services to be deployed crosstalk can have a major impact on performance.

There are two main ways to deal with the problem of crosstalk. One is echo cancellation and the other is Frequency Division Multiplexing(FDM). Echo cancellation is where the signals are transmitted and received using the same frequency. The transmitted waveform is familiar and can be accounted for by the attenuated receiving waveform. FDM is where the signals are transmitted and received at different frequencies. This is effective because it eliminates NEXT because the transmitted signals are of a different frequency than the signals being received over adjacent loops in the cable bundle. FDM based transmission systems often offer better performance in relation to echo-cancellation systems when it comes to eliminating crosstalk. The disadvantage of FDM is that it requires the use of higher frequencies than echo-cancellation technique, resulting in shorter distance that a signal can be transmitted using FDM.

The significance of these line codes is that line codes must be adopted as standards for the industry to ensure interoperability. Right now in many cases Dsl modems are only compatible with the modems of that manufacturer. Standards ensure that hardware is compatible throughout the industry.

When deciding on how to best deploy Dsl in the local loop, the telcos must investigate the characteristics of the particalar line and the characteristics of the other loops in a bundle. Attenuation will cause even more problems in the local loop as higher and higher speeds are desired. Higher frequencies will be required, thereby shortening the length of local loop the signal can traverse.

There are several variations of Dsl that have been developed since the early 1990's. There are two major differences among Dsls. Dsls differ in that they can offer different data rates dependent upon line length. The other difference being that Dsls can have a symmetric or asymmetric data flow. Symmetric means that data rates are the same in both directions. While asymmetric means that data rates are higher from the CO to the subscriber than from subscriber to CO. With this in mind let's have a look at the different types of Dsls available and their applications.

HDSL(High bit rate Dsl) was the first Dsl to come about. HDSL was developed as a way to provide repeaterless T1 and E1 connections. The traditional method called for using 2B1Q line code to provision 1.544 Mbits/s over a single repeatered line. The new method called for using two lines transmitting 7.84 kbits/s. By reducing the speed at which signals need to be sent on each line, the frequency required to transmit the signals was lowered, which lengthens the length of loop over which the signal can be sent. By using HDSL as a way to provide repeaterless T1 connections line length approaches 12,000 ft.

The next form of Dsl is the SDSL, which stands for Single or Symmetric Dsl. SDSL offers the same service as HDSL, but over a single wire pair. This can be done at loop lengths of around 10,000 ft. SDSL gives up some distance to HDSL by using higher frequencies to transmit signals.
Studies were performed on the loop itself and it was discovered that signals could be sent a greater distance from the CO to the subscriber than from the subscriber to the CO. This is because crosstalk is more prevalent at the CO than at the subscriber end of the loop. Cables are bundled together at the CO and there is more chance for interference. At the remote subscriber the cables have branched off and crosstalk is less of a problem.

Provisioning longer loop lengths by providing asymmetrical services, led to the development of ADSL, or asymmetric Dsl. ADSL was originally developed with hopes of providing video on demand, interactive t.v. and POTS(telephone service) over the same loop. To provide video on demand it was determined that a continuous bit rate of 6 Mbits/s was necessary to ensure picture quality. Several line codes were tested to determine which one could effectively provide this rate. Three main codes were capable, including CAP, which was seen earlier as a way to provision HDSL. The standard that was adopted, though, was called DMT, or descrete multitone. In the process of the trials, it was realized that this asymmetric data rate was perfect for providing Internet access.
This led to the adoption of asymmetric data rates for Internet access.

This is what is becoming available today in the form of RADSL, or Rate Adaptive Dsl. RADSL is a rate adaptive variation of ADSL. This is a great advantage because data rates can be adjusted depending upon line length and conditions.

G-lite is the latest version of Dsl. It is really just another version of ADSL. G-lite offers more flexibility in that it can be run at lower speeds, therefore G-lite uses lower frequencies, extending the distance it can effectively travel in the loop. Another bonus of G-lite is that it is easily installed and set up. This is one drawback of other forms of Dsl. Dsl services usually require the knowledge of a technician to set up the modems and other hardware. This is very significant in making this Dsl widely available to consumers.

It looks like, for now anyway, RADSL and G-lite are the most likely to receive the most use for connection to the internet because they are the most compatible with existing local loop conditiions. HDSL will likely remain prominent as a way to provision T1 connections.

Digital Subscriber Line	Line Length and Data Rate	Data Flow	Possible Applications
HDSL-High Bit Rate Dsl	1.544 Mbits/s over 2 pairs for T1, 3 for T2	Symmetric up to 12,000 ft.	Provision T1 connection, PBX network connections, Internet servers, and private data networks
SDSL-Symmetric Dsl	1.544 Mbits/s over single pair	Symmetric up to 10,000 ft.	residential video conferencing, and remote lan access
ADSL-Asymmetric Dsl	1.5 to 8 Mbits/s down, 16 to 640 kbits/s up	Asymmetric up to 18,000 ft.	Video on demand, interactive t.v. Internet access
RADSL-Rate Adaptive Dsl	up to 7 Mits/s down, 600kbits/up	Line length depends on rate, symmetric or asymmetric	All of the above
G-lite	up to 1Mbit/s down, up to 200 kbits	asymmetric	primarily Internet access

Although just about every major telephone company is running a trial or is beginning to make Dsl widely available, prices need to come down. There are also a lot of bugs that the telcos need to work out. Because Dsl services are new, the telcos haven't exactly worked out all the issues that need to be attended to, in order to smoothly provide Dsl. This would cover everything from billing to installation. Examples of experiences can be found in the Usenet newsgroup, comp.dcom.xdsl. Many individuals are complaining that in order to get connected using Dsl services, they had to go through a series of problems before they could get their service running properly. However, One would have to believe that as telcos gain more experience and variations of Dsl become easier to set up, these problems will become less of an issue.

Here in Austin, SWbell is running a Dsl trial called Fastrac Dsl. This program is barely off the ground. When I called, I was told that in my area I could get non-dedicated service, which is dial up, at 384 kbits/s for $150 and 1.5 mbits/s for $250. Dedicated, which is a constant connection, costs $395 and $995. There are also additional costs involving line installation. Jump Point, a local Internet service provider, would have to provide the Internet service which would cost an additional $95 to $145 for non-dedicated and $395 and $995 for dedicated service. These prices are way to high to for the home user to consider. Dsl services in Austin are not really ready for deployment at this time.

U.S. West is offering Dsl services in 40 cities throughout 13 Midwestern states. They have a package called megabit services, which is much more affordable than the Fastrac Dsl. This service offers 256 kbits/s at $59.95 and $110 for installation, if you do it yourself. Internet service will also need to be purchased. This is still a little high, but it is getting closer to affordable.

Before Dsl services become widely available and affordable, there needs to be more competition among local network service providers. Right now the incumbent local exchange carriers, such as SWbell, control the local loops. Incumbent local exchange carriers, or ILECs, make it hard for other Network Service Providers to enter the local phone service market. Until more competition is introduced and other companies can afford to use the local loop for provisioning local phone service, Dsl services won't be widely available at affordable prices.

Dsl services, themselves, appear to be coming on as a viable way to connect to the internet. Services, such as G-lite, promise to be widely implementible over the local loop. The developers of Dsl services appear to have bridged the gap between the local CO and the subscriber. The local loop can be used for megabit services. Now we just have to wait until it becomes affordable.