Telephone companies put great efforts into providing digital data communication over the existing copper wire networks. The reuse of existing copper saves telephony companies the expenses of upgrading millions of copper wiring already in place. The provision of ADSL services simply requires a pair of ADSL modems between each customer's premise and the exchange building, additional concentrators in the exchange building, and some wiring and equipment update. However, ADSL is not the sole technology being promoted for high-speed data communication. The success of ADSL will lie on the telephone industry's ability to resolve some of the technical and non-technical issues discussed in this paper.
Within next several months, BC Tel will start offering an Asymmetric Digital Subscriber Line (ADSL) service to British Columbians. ADSL is a new modem technology that uses the ordinary copper-wire telephone lines to give users high-bit-rate data communication services. This technology will save telephony companies the expenses of upgrading millions of copper wiring already in place. The drastic data rate increase makes many bandwidth-hungered applications available to average home users. Many people believe that ADSL, within a couple of year time, will be matured and massively deployed. Some people, on the other hand, believe that ADSL is "dead in its tracks" [1] with the problems of complexity, standardization, and many other issues. Besides facing various technical issues, ADSL must also compete with some other new technologies such as Hybrid Fiber Coax (HFC), which is the ADSL's counterpart in the Cable TV world. This paper will describe ADSL in general and discuss its future when considering various issues.
ADSL is one of the newer DSL technologies pursued by telephone companies. The intent of DSL technology is to deliver a high performance and cost -effective way for transmitting at high speed over existing copper wires, without the need for repeaters or special line conditioning [2]. An ADSL line consists of a pair of telephone wires and a pair of ADSL modems, with one modem installed in the telephone exchange building, and the other installed in the user's premise. Figure 1 shows an ADSL connection described in [3].
Figure 1. An ADSL Connection
The 4 kHz bandwidth of the telephone pair is divided into a unidirectional high data rate channel to customer, a lower speed bi-directional control channel, and a channel for plain old telephone service (POTS). Theoretically, ADSL is able to deliver at a maximum speed of 8.448 Mbps [4]. Currently, many telephone companies are able to provide a 1.5 Mbps to 6 Mbps data rate to their customers, depending the distance between the customer's premise and the exchange building. Table 1 shows the practical limits on data rate in one direction compared to line length of 24 gauge twisted pair [5]. The bi-directional channel is used for transporting programs selections and other interactive controls in the network.
Table 1. Practical Limits on Data Rate Compared to Line Length
| Data Rate Limit | Line Length | |
| DS1 (T1) | 1.544 Mbps | 18,000 feet |
| E1* | 2.084 Mbps | 16,000 feet |
| DS2 | 6.312 Mbps | 12,000 feet |
| E2* | 8.448 Mbps | 9,000 feet |
| *speeds available in Europe | ||
ADSL modems create multiple channels by dividing the 4 kHz bandwidth of a telephone line in either one of the two ways - Frequency Division Multiplexing (FDM) or Echo Cancellation. FDM assigns one band for delivering downstream data and another band for transporting bi-directional control signals. Echo cancellation assigns upstream and downstream bands in such a way that they overlap each other and separate the two by means of local echo cancellation [6].
ADSL modems also provide error correction by aggregating data from the upstream and downstream into blocks and attaching an error correction code to each of the blocks.
An ADSL modem interfaces a computer with the external network by converting digital data to RF signals to be transmitted across the telephone line and retrieving the digital data from RF signals received from the peer ADSL modem. The connection between the ADSL modem and the computer, however, will not be via the computer serial port because the serial port cannot support the increased data rate. Instead, most computers have a network interface card (NIC) that connects to the ADSL modem's 10Base-T Ethernet port. This configuration makes connections to ADSL modems simple since it does not require special protocols or drivers. Any standard Ethernet network interface card will work [5]. Computers can view the external network as parts of an Ethernet network. Figure 2 shows an example of the Ethernet access network configuration.Figure 2. Ethernet Access Network Configuration
An ADSL modem pair, as mentioned above, converts digital data from the computer to RF signals and converts RF signals received digital data. Figure 3 shows the interior of an ADSL modem.Figure 3. Interior of an ADSL modem
Currently two transceiver technologies are available for developing ADSL products. One is Carrierless Amplitude/Phase Modulation (CAP), and the other one is Discrete Multi-tone (DMT). Each of these methods has its strengths and weaknesses. They both handle noise well and can operate in older copper plant [7]. DMT can dynamically adapt to changing noise environments by switching to frequencies with less interference. DMT is rate adaptive. It matches the modem's rate with the rate of the application. DMT is also the modulation technique specified in the ANSI T1E1.4 standard. Although DMT should be the more preferred technology for implementing ADSL products, it is not true for the time being because of its high cost. CAP was used in the first commercially deployed ADSL transceiver, and it is installed on more lines than any other ADSL technology. It is currently the most cost-effective way to implement ADSL products. Although the cost of DMT is expected to decline, CAP should still be the preferred choice in the next couple of years.
The original driving force behind ADSL was to provide Video On Demand services. Telephone companies, hoping to get their shares in this thought to be enormous market, looked for ways to allow them to deliver high data rate video services. They could either replace their entire infrastructure with fiber, or they can stay with their copper. Cost and revenues had led telephone companies to implement a sequence of broadband access network topologies that end with existing copper stretched between exchange buildings or remote nodes and customer premises [8]. One of the broadband access network topologies is of course ADSL. Unfortunately, VOD did not boom as people had expected. Instead, the Internet dominates public discussion of the Information Superhighway. The Internet has changed from text-based to currently multi-media based with all sorts of graphics, audio, and videos. The public wants to have video conferencing, long -distance phone calling, and even video on demand over the Internet. These services require high bandwidth. Another application that requires high data rate communication is remote LAN access. With many large companies downsizing and their office locations become more geographically distributed, their employees need to access the company servers from outside the office. Will the telephone companies be able to use ADSL to meet these demands? Will they be able to capture parts of the market at all? No one is able to answer these questions now, but the success or failure of ADSL will be greatly dependent on the factors discussed below.
Obviously, ADSL is not the only technology that can provide higher data rate communication than those currently available. Of various ADSL's competitors, Hybrid Fiber Coax (HFC) from cable companies stands out from the rest. When comparing ADSL and HFC, one technology is not better than the other, at least not right now. Both technologies are still a couple of years away from maturity. They all have their strength and weakness. HFC can provide a theoretical speed of 30 Mbps, which is about 3x that of ADSL. Cable modem rates do not depend upon coaxial cable distance, as amplifiers in the cable network boost signal power sufficiently to give every user enough bandwidth [9]. Unfortunately, this 30 Mbps bandwidth is shared by tens of other users; as more users are online, the capacity available to one user drops. Ingress noise also affects the bandwidth available to users. HFC may offer a less expensive network solution because of its shared architecture. However, it does have high infrastructure costs because of the required upgrades of existing networks. Details on HFC can be found in [9], [10], [11], [12].
Telephone companies reuse billions of miles of existing copper wires along with ADSL to provide high data rate communication services. The success of ADSL and the telephone's ability to capture market shares is of course lies on the availability of telephone lines. According to Maxwell [8], there are almost 700 million copper access lines in the world today; there will be about 900 million by year 2001. Those businesses, homes, schools that may need high-speed networks have already wired. Once the required equipment is installed, high-speed communication will be as simple as telephone communication today.
The cost of the ADSL technology can be separated into two categories: the cost for implementing the infrastructure; that is, the cost paid by telephone companies; and the cost for setting up and maintaining an ADSL line at the customer's premise; that is, the cost paid by a customer. In order to provide ADSL services, telephone companies need to upgrade their exchange buildings with ADSL modems and ADSL concentrators. Although most of the copper wires need not be replaced, some older copper wires and telecommunications equipment need to be updated to provide better services. As an example, BC Telecom have invested CDN$577 million in 1996 to update their telecommunications equipment and infrastructure, as well as improve telephone service [13]. As mentioned above, the equipment needed by an ADSL customer include a telephone line, an Ethernet network interface card (NIC), an ADSL modem, and an ADSL service package from the local telephone company. The price of a network interface card is around CDN$150. An ADSL modem currently costs about $CDN1500. However, many telephone companies are supplying modems to their subscribers as part of their ADSL service. A monthly subscription fee of $50 to $100 will probably include an ADSL modem and an ADSL service package with about 100 hours of Internet access. Although the price of an ADSL is currently in the $1000 range, people believe that the price will drop as full ADSL deployment is being carried out in many parts of the world. Briere, Telechoice president, said that ADSL will migrate from being a standalone piece of equipment to integration within both telephone company and data networking equipment, a process that would drive costs down still further [14]. The currently commercially available ADSL modems are developed using chipsets from AT&T Paradyne, Amati, and a partnership of Analog Devices and Aware Inc. These manufacturers believe that a single-chip-set ADSL solution will soon be available. The cost of such a chip set will be lowered than those currently available [14], [15].
Even though ADSL is a miracle inside the copper wire, it is not without problems. Although ADSL can theoretically deliver up to 6 Mbps, many of the existing copper wiring cannot support such a rate. For example, loading coils, which were originally installed on long circuit loop to cancel noise, low-pass filter transmitted signal to 4 kHz. ADSL modems, of course, require bandwidth higher than 4 kHz. Thus, loading coils existed between the exchange offices and customers' premises must be removed. In addition, telephone networks are plagued by overlong loops, which attenuate signals; and bridged taps, which are the unterminated pairs. As mentioned above, an ADSL line can support a data rate of 1.544 Mbps when the distance between the customer's premise and the exchange building is less than 18,000 feet. Shorter local loop length is needed if a data rate of 6 Mbps is desired. The ADSL Forum estimated that about 80 percent of the phone customers in the U.S. are within 18,000 feet of the exchange buildings [4]. Performance of an ADSL line is greatly affected by heat dissipation, power consumption, and crosstalk. Crosstalk is the noise dependent on transmitted signal. Near end crosstalk (NEXT), which is the interference into a receiver from a transmitter at the same end of the cable and far end crosstalk (FEXT), which is the interference into a receiver from a transmitter at the opposite end of the cable, are two types of crosstalk [3] (Figure 4).Figure 4. FEXT and NEXT
The effect of FEXT on transmitted signals is usually small when compared to NEXT and is usually ignored. NEXT power rises monotonically as the wire length increases. It quickly approaches a constant asymptote [3]. Crosstalk appears because a phone line is not a continuous, unbroken pair of wires that run from the exchange building to the customer's premise.
Another technical issue related to ADSL is the deployment of ATM. Current ADSL configuration is for deployment in the IP world. ADSL modems cannot handle non-IP traffic. It is inconvenient for connecting multiple service providers into the same access network [5]. Most of all, ADSL today is not a switched technology. However, many telephone companies still believe ATM is the future. If that is the case, ADSL vendors must find a way to incorporate ATM switching as soon as possible. Unfortunately, ATM can not be economically implemented today because ATM network interface cards are still expensive and problems with ATM signaling and conversion have not been worked out yet. In addition, the switching to ATM from IP would lead to a more complex access infrastructure. As the infrastructure becomes more complex, the more likely ADSL will be delayed.
Two major issues that will affect the acceptance of ADSL are standardization and interoperability. The two issues are related to each other. As mentioned above, currently two transceiver technologies are available for developing ADSL products. CAP (Carrierless Amplitude/Phase Modulation) has already been used in many of the existing ADSL products. However, ANSI has specified DMT (Discrete Multi-tone) to be the modulation technique in ANSI T1E1.4 (ETSI has not yet specified a standard). Theoretically, DMT is more susceptible to noise and can support longer wire lengths because it is noise adaptive and rate adaptive. Practically, however, the two specifications are not too far apart. Currently, ADSL modems that employed CAP are not compatible with modems those employed DMT. Thus, customers buying ADSL modems must check with their local telephone company to make sure they have the appropriate equipment. Fortunately, many of the telephone companies are currently supplying to customers ADSL modems as part of their ADSL service package. Thus, customers do not have to worry about interoperability.
Another issue that will affect ADSL deployment is manageability. "What kind of network management is available?" "Can the network be managed down to the element level?" "Will there be a single system that can mange products from different vendors?" These are just a few of the questions raised when a customer decides to employ ADSL. Each vendor will probably supply management software for managing its products. However, the question is, will any interoperable management systems be available to manage products from multi-vendors? ADSL will obviously not going to succeed if a network cannot be easily managed from the system level down to element level.
The ADSL technology extends copper infrastructure into the 21st century. It enables telephony companies to provide high data rate communication services over the existing copper twisted pairs. The major market focus for ADSL is to provide Internet access and remote LAN access, where customers are willing and capable of paying. Several issues will destined the future of ADSL: the competitiveness of ADSL against other technologies, the cost for implementing ADSL infrastructure, the cost for employing ADSL in customers' premises, the influence of ADSL's performance by crosstalk, heat dissipation, and power consumption, the needs for upgrading ancient copper wiring, the strategy for incorporating ATM into the current infrastructure, standardization and interoperability, and the ease of managing ADSL networks. Telephone companies are working very hard to try to address most of these issues. However, will they be able to resolved these issues in time? Or will the ISDN scenario occur once again? Telephone companies will probably have to wait for another year or two before they will know if their efforts get paid off. In the mean time, they better work hard to stay as competitive as possible.