To illustrate how DSL works, we’ll use the example of an ADSL connection from your home to your service provider’s central office (CO).

In DSL, voice and data get transferred simultaneously over your existing twisted-pair copper telephone lines by using different frequency ranges on the same line. Voice is transferred on lower frequency bands and data on higher ones.

The technology to do this resides in the DSL transceiver or modem that’s installed both at your end and at the end of your service provider. A DSL modem on your end sends data over the telephone line to your service provider’s CO. At the CO, a DSL Access Multiplexer (DSLAM) terminates and aggregates incoming ADSL lines. It redirects the voice traffic to the public switched telephone network (PSTN) and the data to a high-speed digital line that connects to the Internet. Let’s look at the nitty-gritty of the process.

The DSL modem on your end divides the available bandwidth of a telephone line using either frequency division multiplexing (FDM) or echo cancellation. In FDM, one frequency band is assigned for upstream data and another one for downstream data. The downstream path is then divided into high-speed and low-speed channels, and the upstream path into low-speed channels. In echo cancellation, the upstream path overlaps the downstream path and the two are separated by a method called local echo cancellation. For both these methods to work, you need to install low-pass filters or splitters, called POTS (Plain Old Telephony Service) splitters, with your DSL modem. These separate low frequency voice signals from high frequency data signals, so that one doesn’t interfere with the other, and you get simultaneous access to telephone and Internet services. Such a splitter would be installed at the CO too. Since human voice can be transmitted below a frequency of 4 kHz, most low-pass filters block access above 4 kHz.

Transmitting data with DSL

Transmitting digital data over an analog (telephone) line is a complex process, and various problems can arise. In copper lines, distortion is higher on higher frequencies, and since digital data requires higher frequencies, transmitting it is a challenge. There are also problems like thermal noise, crosstalk (interference between nearby cables), and attenuation (signal loss because signal power diminishes as it travels across a medium, especially for long distances). DSL modems use a process called modulation to address some of these problems. Modulation also enables the transfer of large amounts of digital data, which is what makes DSL a high-bandwidth solution.

Simply speaking, modulation is a method by which a data signal is transferred from one point to another over long distance. The data signal is first put on top of what’s called the carrier signal, which is stronger either in amplitude or frequency. The resulting encoded signal is then sent and recovered at the receiving end by a process called demodulation. In DSL, the message signal from a sending modem modifies a high-frequency carrier signal to form a modulated wave. At the CO, the receiving modem demodulates this signal to recover the data.

DSL uses either Carrierless Amplitude Phase (CAP) or Discrete MultiTone (DMT) to modify the carrier signal. Both of these use the same modulation technique—Quadrature Amplitude Modulation (QAM)—but implement it in different ways. Let’s take a look at how they’re implemented.

In QAM, two independent message signals are used to modulate two carrier signals that have the same frequencies, but different amplitude and phase states. This enables bandwidth conservation by allowing two digital carrier signals to get transmitted on the same bandwidth. QAM receivers can make out whether to use lower or higher numbers of amplitude and phase states to overcome noise and interference during transmission.

CAP uses a slightly complicated procedure, in which the carrier signal is suppressed before transmission.

So the message signal is first modulated by a carrier signal and stored in memory. Then, pieces of this modulated signal are reassembled and passed through a band-shaping filter before being transmitted. The band-shaping filter actually imposes a carrier on this assembled signal, converting it into a modulated wave. The advantage with CAP is that it has lower peak-to-average signal power ratio relative to DMT. So its end equipment requires lesser power than DMT. CAP also tests the quality of the access line before transmitting and implements the most efficient version of QAM so as to minimize signal loss during transmission.

DMT divides the available carrier frequencies into 256 discrete sub-channels or tones, and checks for the carrying capacity of each sub-channel before transmission. Data is then divided into bits and distributed to sub-channels depending on their ability to carry the transmission. Because higher-frequency signals on copper lines suffer more loss due to noise than lower-frequency ones, more data is sent on lower frequencies than on higher ones.

Between CAP and DMT, CAP is less expensive, but is not an industry standard. DMT is an industry standard supported by American National Standards Institute (ANSI), European Telecom Standards Institute (ETSI), and International Telecommunications Union (ITU). A variant of DMT, called DWMT or Discrete Wavelet MultiTone, is under development. This isolates the sub-channels even further. Once fully developed, it may become the de facto ADSL protocol for long-distance transmission, especially where high interference is prevalent. Other versions of DMT, like Synchronized DMT and Zipper are being proposed for use with VDSL.

DSL is being implemented in various parts of the world and is becoming a popular choice for providing content like multimedia communications, video-on-demand, Internet and intranet access, and remote LAN access on your PC. However, there are issues like interoperability of various xDSL technologies, interference between different services in the same binder group (a collection of twisted pair wires that share a common ‘sheath’) which can result in degradation of nearby signals that need to be resolved before xDSL technologies can be widely deployed.

Pragya Madan