Introduction
The telephone network is a ubiquitous and mature medium of communication with high levels of consumer penetration and potential data carrying capacity. DSL (Digital Subscriber Line) is one of the technologies used to provision high speed links to data networks over twisted pair Public Switched Telephone Networks (PSTN). These networks were originally designed to carry voice traffic over twisted pair copper lines (Jones, 2006).
The Line in DSL refers to link between the consumer’s location and the local exchange. The PSTNs usually operate in the frequency range between 300Hz to 3.4 kHz of the total 1.1 kHz of spectrum available for voice communications. DSL usually provides high speed data communication by operating in the unused frequency range of the telephony spectrum above 3.4 kHz. DSL technologies support both bus and star topologies of network deployment and utilize pairs of compatible transceivers to facilitate communication between the consumer site and the telephone exchange (Littman, 2002).
The transceivers usually installed at telephone exchanges or central offices must be able to separate the voice and data traffic, the voice traffic is usually sent over the PSTNs while the data traffic may be forwarded to high speed data networks. The telephone exchanges provide services to subscribers using DSLAMs (Digital Subscriber Line Access Multiplexer). In the upstream direction the DSLAM multiplexes traffic from multiple user ports carrying traffic of a single user to a few network ports carrying traffic for multiple users. In the downstream direction DSLAM de-multiplex traffic from network ports onto the correct user ports (Jones, 2006; Jayant, 2005; Golden, 2007; Maxwell, 1999; Smith, 2007).
The range of DSL technologies is quite broad, but they can be roughly broken down into two types.
Symmetric DSL
In these types of technologies both the upstream and downstream links support the same data transfer rate.
Asymmetric DSL
In these types of technologies the data rates of upstream and downstream links are dissimilar.
The above description is not complete since some technologies support both symmetric and asymmetric modes of communication and variable transfer rates (Jones, 2006).
Transmission Technologies
Before delving too deep into details of DSL technologies a review of transmission technologies deployed over voice networks is in order. The telephone voice data network, commonly referred to as Public Switched Telephone Network (PSTN) operated in the frequency ranges of 300 Hz to 3.4 kHz. Voice signals are primarily analog in nature; however advances in modern technologies have led to digitization of these analog signals to digital signals. Digitization of signals usually takes place in the main backbone networks connecting Company Offices (COs). The voice signal is carried in analog form from the user’s location to Company Offices (COs) using a copper twisted pair also referred to as a local loop.
The analog signal is converted to digital form at the Company Offices (COs) by a codec (coder/decoder). This is followed by transmission of signals in digital form over backbone networks to the Company Office (CO) at the other end. Once transmitted the signal is converted back to analog form by a codec (coder/decoder). After conversion the aforementioned signal is transferred in analog form to the subscriber’s location by transmitting it over the local loop. Modems (Modulators/Demodulators) were introduced in the 1950s for transmission of data traffic over voice networks. The earliest modems were designed to achieve speeds of 300 bits/sec using Frequency Shift Keying (FSK) modulation.
With the evolution of technology higher bit rates and full duplex transmissions became possible. The V.22 standard is a case in point. Data rates of 1200 bit/sec were achievable at the start, which were later extended to 2400 bit/sec. Subsequent technology developments led to formation of standards capable of achieving comparatively higher data transmission rates of up to 33.6 kbps with various fallback options. Pulse Coded Modulation (PCM) modems commonly referred to as V.90 modems were developed later, which support downstream rates of up to 56 kbps downstream, that is from the CO to the subscriber’s location. These projected data transfer rates are based on the assumption of a digital path being available from source of data to the CO. Usually this is not a problem since most ISPs offering dial up services have direct connections to the Public Switched Telephone Network (PSTN).
The V.90 modems provide an upstream data transfer rate of 33.6 kbps based on the V.34 standard. Voice modems have been continuing to push the limits of technologies however there is a limit to bandwidths that can be achieved using voice band technologies. Raising demands on part of the users for higher bandwidths to support increasingly content rich media have led to creation of other technologies that support higher bandwidths and data transmission rates over exiting Public Switched Telephone Networks (PSTNs). The data access patterns for most residential users are asymmetric in nature, requiring availability of different levels of bandwidths in the upstream and downstream direction.
Usually common activities like web browsing involve higher transfer of data from the source to the destination, i.e. from Central Offices (COs) to subscriber locations. Business user’s data access patterns differ from those of home users. Business users use data in a symmetric pattern with equal amounts of data being transferred both upstream and downstream. Increasing customer requirements have led to the development of newer technologies like DSL (Jones, 2006; Tanenbaum, 2003).
DSL Technologies
The various common DSL technologies are described as follows
Basic Rate ISDN
The BRI ISDN belongs to symmetric (capable of transmitting at the same data rate both upstream and downstream) class of DSL technologies. It can operate at 160 kbps over a distance of approximately 5.5 Km. ISDN comprises of two channels operating at 64 kbps for voice or non voice traffic and a single data channel operating at 16 kbps for signaling, control and data packets (Jones, 2006).
Asymmetric Digital Subscriber Line (ADSL)
As the name implies ADSL belongs to the asymmetric (capable of transmitting at different data rates upstream and downstream) class of DSL technologies. ADSL uses a pair of modems, one situated at the client’s site while the other is situated at the service provider end like the local telephone exchange or central office. The term ADSL Terminal Unit-Remote (ATU-R) is used for the modem or transceiver situated at the client’s end while the term ADSL Terminal Unit-Central Office (ATU-C) is used for the modem situated at the local telephone exchange or central office. At the client’s location ATU-Rs perform modulation and demodulation of signals for optimum transmission over copper wires.
The ATU-Rs work in conjunction with hardware equipped with NICs (Network Interface Cards) to transmit voice and data signals over PSTN to provider locations. ATU-Rs use error correction schemes like forward error correction where redundant data is added to the messages to enable the receiver to perform error checking without asking the sender for more data and echo cancellation using echo suppressers. This reduces any disruptions that might occur due to noise at the PSTN level (Littman, 2002).
The previously mentioned DSLAMs and ADSL transceivers situated at telephone exchanges and central offices are classified as ATU-Cs (ADSL Terminal Unit-Central Office). Hubs, bridges and routers located at central offices charged with redirecting voice calls to PSTN networks and data, voice and video to IP networks are also classified as ATU-Cs. The DSL capable technologies used to implement backbone infrastructure connecting various central offices include ATM, SONET/SDH, IP, Fiber Channel, Frame Relay and Gigabit Ethernet (Littman, 2002).
ADSL uses various techniques to determine the optimum bit transfer for the given transmission conditions. Thus ADSL is a variable bit rate asymmetric transmission DSL technology. The upstream rates in ADSL are lower as compared to downstream rates, thus downstream communication utilizes more bandwidth than upstream communication. The frequency range for upstream transmission lies between 25-138 kHz. In the case of downstream transmission, utilized frequency ranges depend on the mode of operation of transceivers. Full overlap between upstream and downstream bands is used in circumstances where larger amounts of bandwidth are required.
However this might lead to line noise and sophisticated echo canceling techniques need to be implemented for noise reduction. Full overlap between upstream and downstream bands utilized frequencies in the range between 25 kHz to 1.14 MHz. In the case of Frequency Division Duplex (FDD) non overlapping downstream and upstream bands are used and they utilized frequencies starting from 138 kHz for downstream communication (Jones, 2006; Stallings, 1998).
Splitter less ADSL
Splitter less ADSL was another standard proposed by ITU to ease installation of ADSL services at client premises, by removing the need to install splitters at client end. Use of a modular low pass filter in series with the telephone is required to prevent interference from Plain Old Telephone service (POTS). ADSL users need to be with 18000 feet of the telephone to avoid experiencing a degradation of service. ADSL service quality also depends on the quality of the local loop (Jones, 2006).
HDSL 1
HDSL 1 is a symmetric transmission DSL technology, providing same bit rates for upstream and downstream communication. HDSL allows data transfer rates of up to 384 Kbps over a single twisted pair copper cable. Usually more than one twisted pair copper cables are used to achieve higher transmission rates up to 2.048 Mbps. The increased usage of T1 and E1 connections at consumer locations led to the development of this technology as an alternative to traditional Alternate Mark Inversion (AMI) and High Density Bipolar (HDB3) modulation based technologies which were problematic due to excessive crosstalk.
The method used for data compression in DSL is 2B1Q (Two Binary, One Quaternary) line code modulation. The effective range of HDSL1 transmissions without the use of repeaters for signal regeneration lies between 12000 and 15000 feet. The use of more than one pairs of twisted pair copper cable helps in increasing reach because the bandwidth used on each pair is lesser as compared to a single twisted pair cable. This leads to lesser energy losses (attenuation) per km of transmission. Attenuation is frequency dependent, the lesser bandwidth availability per twisted pair cable results in reduction of attenuation during transmission.
Lower bandwidth also helps ensure compatibility at a spectrum level with existing systems. The downside of using more than one pairs per customer is that fewer customers can be served with a given number of twisted pair; single pair cables have the advantage of serving more customers while sacrificing some level of spectral compatibility and suffering losses due to attenuation. This led to the creation of high performance DSL which used single pair cable while using a smaller bandwidth at the same time (Jones, 2006; Littman, 2002).
HDSL 2
The need to retain spectral compatibility with existing systems while achieving acceptable reach over a single twisted pair copper cable led to the development of HDSL2 (High Bit Rate Digital Subscriber Line Phase 2). Like HDSL HDSL2 is based on PAM but it provides several enhancements that were not available with HDSL. One of the added features offered by HDSL2 is the use of error correcting codes.
This implies that HDSL2 sends enough redundant information with each block of data sent to enable the receiver to work out the errors without the need to retransmit data. HDSL2 also uses Trellis-Coded Modulation (TCM) which allows for several decibels of extra performance to the system. TCM is a standard for high speed modems that allows for automatic error correction using extra bits in each sample. One consequence of using TCM is that the signal is equalized before transmission by the transmitter which is in direct contrast to the traditional method of equalization at the receiver end.
The performance requirements for HDSL2 stipulate operation in the presence of various disturbances according to carrier service area (CSA) requirements. In simple words 2.7km over 0.4 mm wires and 3.7 km over 0.5 mm wire. HDSL2 can operate over longer distances with the use of repeaters however repeater use causes problems due to spectral incompatibility. This problem was the reason behind creation of HDSL4 (Jones, 2006).
HDSL 4
The HDSL4 standard was created in order to facilitate operation over long distances without the use of repeaters and thus maintaining spectral compatibility. This is achieved by using two paired cables for transmission of data. The HDSL4 standard shares the principle of transfer of half of the signal over each paired cable to reduce bandwidth usage and hence reduce attenuation. The underlying technology shares some of the features of HDSL2 like coded 16-PAM modulation with different spectrums of frequencies available for upstream and downstream communication (Jones, 2006).
Symmetric High Bit Rate DSL (SHDSL)
SHDSL is a symmetric DSL technology supporting identical bit rates both upstream and downstream. It is ideal for applications requiring high bandwidth usage. These applications include Tele-entertainment, corporate LAN connectivity, video conferencing etc. SHDSL can be used to deliver multimedia to the customer’s location and can serve as a vital link between the optical fiber termination point for FTTN (Fiber to the Network) and the subscriber’s home (Littman, 2002).
The Holy Grail for user connectivity is FTTH (Fiber to the Home). However this involves significant technical and financial challenges and has remained largely unattainable. However some progress to this effect has been made in the form of intermediate architectures like FTTN. FTTN involves transmission of data to local neighborhood network points over high speed fiber just reducing the distance required for subscribers to connect to high fiber networks (Jones, 2006).
These points can be connected to the subscriber’s location using SHDSL which lends itself well to the projected high bandwidth demand of home and office users, and enables usage of bandwidth hungry multimedia applications. SHDSL is also known as G.SHDSL. A common feature between SHDSL and HDSL2 is that both support duplex rates at 1.54 Mbps and 2.048 Mbps. SHDSL works well with ATM technology and supports the G.HS protocol for communication between compatible DSL devices (Jones, 2006).
Symmetrical or Single Line DSL
SDSL is another popular DSL technology for providing multimedia services that require same upstream and downstream rates. It is used primarily for IP Telephony, Web Hosting, Tele Banking, Tele Working and Video Conferencing. SDSL deployments are not governed by any standards and are propriety in nature (Littman, 2002).
Very High Bit Rate DSL (VDSL)
VDSL is currently the first version of DSL technologies. VDSL is usually used with FTTN and FTTC which enable the transmission of data to local Optical Network Units (ONUs) over fiber. This data is then transmitted to the subscriber location over short distances using VDSL. VDSL enables such bandwidth hungry broadband applications as multi channel distribution, telephony services, fast internet connectivity, video on demand, high quality video conferencing, high definition television programming (HDTV), telemedicine, E-commerce and Tele education (Littman, 2002). VDSL supports either symmetric or asymmetric transfer rates with first generation VDSL providing either symmetric data transfer rates of 13 or 26 Mb/sec or asymmetric data transfer rates of 52 Mb/sec downstream and 6.4 Mb/sec upstream.
However high data transfer rates can only be maintained over short distances. Bandwidth consumption of VDSL is quite high as compared to ADSL, with VDSL utilizing 12 MHz as compared to 1.1 MHz used by ADSL. VDSL usually uses Frequency Division Duplex (FDD) for bandwidth allocation, which prevents overlap between downstream and upstream communication. Echo cancelled operation, although theoretically possible is not practical due to high bandwidths involved. Asymmetric VDSL is utilized in locations like office towers and apartment complexes (Jones, 2006).
Rate Adaptive Digital Subscriber Line (RADSL)
Installation of equipment usually presupposes a certain level of performance. Devices exist to ensure operation of transmission equipment at required service levels, however variation in service quality in transmission of signals occur due to the quality of transmission media, disturbance, available bandwidth and sensitivity of equipment. Variation in line conditions places a larger onus on sensitivity of installed equipment.
Rate Adaptive Digital Subscriber Line (RADSL) was developed in order to achieve variations in throughput and be sensitive to line conditions. It allows the flexibility to adapt to line conditions and adjust the bit rate at both ends to achieve optimum throughput in each line. Data transfer rates in both upstream and downstream directions can be seen changing with the change in line conditions. Speeds up to 768 kbps are offered by vendors offering RADSL services (Bates, 2000).
DSL Implementation Considerations
DSL technologies offer dependable access to the World Wide Web (WWW), by overcoming traffic grid locks. They enable such technologies as Tele Health, Tele Education, Tele Conferencing, Tele Working, Tele Training etc. Technologies like ATM over ADSL transform the Public Switched Telephone Network (PSTN) into a broadband multimedia network. Downstream rates in DSL transmissions depend on a number of variables including telephone wire condition, thickness and type.
A heavier 24 gauge wire is more helpful in transmitting information as compared to a thinner 26 gauge one. Usually Unshielded Twisted Pair (UTP) cable is recommended for DSL installation. DSL transmissions over local loop are susceptible to loss in signal strength during transmission, dispersion and signal impairment. The signal becomes distorted over large line lengths too. Mixed wire gauges used in coils and telephone lines leads to interference with signal integrity. Faulty equipment at customer’s site, presence of bridge taps and noise generated by wiring at the customer premise all contribute to signal degradation and problems with transmission of DSL signals (Littman, 2002).
The process of planning, deploying and maintaining DSL services of Public Switched Telephone Networks (PSTNs) is quite involved. Any implementation of DSL at a school, office, home or location, requires an understanding of requirements, appreciation of technical and financial risks involved and careful selection of DSL sub technology that meets implementation requirements and goals. Security is one of the chief issues in any form of communication over public networks. DSL uses dedicated virtual circuits which lead to enhanced security as compared to cable modem technologies, which provide point to multipoint shared network connectivity.
Virtual private circuits are used to safeguard communication between the subscriber and the web. Threats due to long duration of connection to the internet are a possibility in DSL technologies and should be mitigated accordingly by using a good firewall and antivirus. Co deployment of various DSL technologies can lead to certain issues in the form of crosstalk, and hence one of the fundamental requirements for new services over DSL is that they should be spectrally compatible with existing technologies. Rules and guidelines exist for co deployment of multiple DSL technologies especially in heterogeneous conditions.
A recent development in DSL technology is Dynamic Spectrum Management (DSM). DSM is designed to allow DSL systems to achieve maximum possible performance while remaining spectrally compatible (Vermillion, 2003; Jones, 2006).
List of References
Bates, R., (2000). Broadband Telecommunications Handbook. New York: McGraw-Hill.
Golden, P. Dedieu, H. & Jacobsen, K., eds., (2007). Implementation and Applications of DSL Technology. Boca Raton, FL: Auerbach Publications.
Jayant, N. ed., (2005). Broadband Last Mile: Access Technologies for Multimedia Communications. Boca Raton , FL: CRC Press.
Jones, E., (2006). Introduction to DSL. In: P. Golden, H. Dedieu & K. Jacobsen, eds. 2006. Fundamentals of DSL Technology. Boca Raton, FL: Auerbach Publications. Ch.5.
Littman, M., (2002). Building Broadband Networks. Boca Raton, FL: CRC Press.
Maxwell, K., (1999). Residential Broadband-An Insiders Guide to Battle for the Last Mile. Somerset, NJ: John Wiley & Sons, Inc.
Smith, R., (2007). Broadband Internet Connections: A User’s Guide to DSL and Cable. Reading, MA: Addison-Wesley.
Stallings, W., (1998). Data and Computer Communications. 5th ed. Upper Saddle River, NJ: Prentice-Hall.
Tanenbaum, A., (2003). Computer Networks. 4th ed. Upper Saddle River, NJ: Prentice-Hall.
Vermillion, W., (2003). End-to-End DSL Architectures. Indianapolis: Cisco Press.