3.1 Basic Performance Enhancement
| A key issue that has limited deployment speeds, especially in the United States where labor is very expensive, is the need in the earliest ADSL systems to send trained field personnel to the customer premises to eliminate a problem that prevents or degrades service. Problems with service may often not be the physical layer and can be related to computer software incompatibilities, external-to-ADSL equipment-interface problems, and so forth. However, as those nonphysical-layer problems have diminished, the issue of unusual noise disturbances or line effects that limit the range become more important. Replacement of the line, or "pair-selection" as it is often called, requires labor and thus increases the cost of ADSL deployment. Thus, most improvement in ADSL has focused simply on the robust delivery of service, that is, to eliminate the need for service technicians to go to the customer. More than one visit to a customer can cause the service to be financially unviable at monthly prices that ADSL consumers can be expected to pay. Splitterless ADSL [1] also attempts to eliminate even the initial visit and is described by customers/providers as a "self-install." Perhaps the most important characteristic of ADSL service is reliability. Customers want DSL service to work well when they need it. Typically, ADSL service is characterized as being "always on" in that the modem physical layer is continuously energized as long as the equipment itself is powered . Thus, no initialization-period time delay annoys the customer. Telephone companies want to achieve this "always-on" connectivity reliably without need for service visits to the customer. To achieve this reliable continuous service, the data rate of ADSL service may be sacrificed. The trade-off then for a given level of good reliability becomes speed of service versus range from a telephone company central office. Figure 3.1 illustrates the classic trade-off between these two parameters. Within the parameter of speed, the level of asymmetry of service becomes important also. Different levels of asymmetry correspond to different ranges for a given downstream data rate. Thus, the ADSL service provider has a difficult compromise to evaluate in attempting to achieve a highly reliable level of speed/service to all customers. Today, the practice is to design for absolute worst-case situations and to guarantee service with little need for visits to the customer. Chapter 11 suggests that such worst-case design is overwhelmingly pessimistic and could be improved to more adaptive automated service provisioning in the future, while still further reducing the frequency of customer service visits. Figure 3.1. ADSL data rates. 3.1.1 Increasing RangeFigure 3.2 illustrates the basic range problem simply. [1] For a given radius around a telephone central office in Figure 3.2, the number of customers served is proportional roughly to the area of the circle. Although it is true that most COs are not at the center of a circle, the basic conclusions inferred from Figure 3.2 will not change substantially. The radius of the circle can be thought of as the length of the longest ADSL loop. As the radius of coverage increases, the ADSL speed decreases because the ADSL signals are attenuated by a longer length of copper wire and thus are more harmed by various noises. One of the noises for an ADSL transmission line is the ADSL upstream crosstalk of other users (or similarly from upstream signals of other symmetric DSLs deployed within the circle). The greater the upstream data rate, the more noise it creates, thus "robbing" data rate from the downstream direction of ADSL. Typically an asymmetry ratio of 8:1 or more doubles the radius that can be achieved (thereby quadrupling the number of customers served without need for visits to those customers) compared with symmetric transmission. However, the actual speed is also important.
Figure 3.2. Illustration of area around central office (CO) with regard to line length. As an example, most (over 90%) U.S. phone lines are believed to conform to what are called RRD rules [1], which basically means they are less than 4 miles in length. [2] Thus, an ADSL system planner might state that at Area = p (4mi) 2 , 90 percent of the customers are served and the other 10 percent will need extra customer service visits to operate properly. Thus, k p (4mi) 2 = .90, meaning at this 4-mile radius, the area of the circle covers 90 percent of the customers ( k is a proportionality constant,
ADSL systems can achieve 1.5 Mbps down and 160 kbps up at 4-mile range in typical situations with splitters and about 500 kbps down and 128 kbps up in worst-case situations without splitters at 4.2 miles. Splitterless operation avoids an initial customer visit for the phone company, but reduces the achievable data rate and reliability, perhaps causing an alternative need for the visit because of service reliability. Thus, the phone company is faced with a classic trade-off. The service provider's choice to date has been to lower the data rate and to use splitterless service to try to ensure a minimum of required service visits, allowing the lowest possible (profitable) per-monthly charges per customer for ADSL service. Although the numbers change in each situation, the reader can appreciate the degree of the problem: increasing 4-mile range just a little bit has a big effect on ADSL service profit. Thus, methods to ensure reliable transmission of higher data rates at the longest ranges have been an initial focus of ADSL system vendors . Some of these methods are described later in this chapter. 3.1.2 Increasing SpeedThe history of data communications and the computer industry has been a steady progression to higher speeds and shows no signs of abatement. Thus, ADSL customers will eventually demand higher speeds. Although 500 kbps service may be initially excellent as a 10x increase over 56 kbps modems, 500 kbps will not remain sufficient in the future. Phone companies anticipate this demand and have invested in programs to shorten phone-line lengths. In the early 1980s, the old Bell System introduced what are called CSA (carrier serving area) design rules, which basically limited new installed phone lines to be less than 2 miles in length. Thirty years of progress with CSA installations has left about 70 percent of the American network within 2 miles of a telephone-company central office or "remote terminal." [3] Southwest Bell Corporation's (SBC) "project Pronto" is probably the best known example of a program designed to bring nearly all telephone lines within 12 kft to increase DSL speeds. This program promises to spend $6 billion over a short period of time (few years) to bring the last approximately 10 million SBC customers (of about 45 million total) within 2 miles of a CO.
A 2-mile radius considerably increases ADSL speeds. In most situations, 6 Mbps down and 600 kbps up is possible. At the very least, with splitterless operation, CSA range means reliable delivery of 1.5 Mbps down and 384 kbps up in telephone-company conservative terms. More generally , yet shorter loops mean even greater speeds, such as discussed in Chapter 7 on VDSL. At least one VDSL draft standard is interoperable with ADSL, so that VDSL offers a mechanism for speeds to increase as phone companies see the opportunity to pay for more fiber and shorter loops. Fiber installation on average is very expensive, and thus its costs need to be shared over a number of subscribers. As line lengths get shorter (i.e., fiber gets longer), fewer customers share the fiber, and so the fiber cost case becomes more difficult. Thus, methods for squeezing more bandwidth on shorter lines are and will continue to be of great interest as customers grow to know and love their DSL service, and thereby start demanding greater bandwidth. It is however, unlikely in the foreseeable future that fiber will be economically deployed directly to the customer on any wide scale. 3.1.3 Improving ReliabilityBoth at the longest ranges and also as speeds increase on shorter loops, reliable service is still most important for telephone companies. Customers expect the service to be available when they need it. Given splitterless installations will increase, which means that ADSL (and any following VDSL in the future) will need to work well in a plethora of different situations so that issues with system performance will eventually return to the reliability of the physical layer transmission line. The remainder of this chapter addresses a variety of effects/issues in maintaining or improving good reliability. Among these are coding methods intended to improve resistance to nonstationary impulse effects and other time-varying line disturbances. Also discussed are analog effects and radio interference issues. |
) Thus, of 1,000,000 customers, then 900,000 would have their premises located within a set of circles that have radius 4 miles from the various central offices servicing all the 1,000,000 customers. Thus also, the remaining 100,000 would be outside the circle and require service visits in this example. If these 1,000,000 customers are to be served at 1.5 Mbps/160 kbps and each customer beyond the 4-mile range requires on the average $1,000 to correct service, then it costs $100M extra in labor to visit/service those million customers, who at $30/month will generate a cumulative revenue of a little over a billion dollars in 3 years . The cost of equipment for DSL service is otherwise about $200/line, so that total equipment cost is $200M. One can see that the extra $100M in labor is significant. If the number of "problem" customer loops is to be reduced from 10 percent in this example to 1 percent, then only 10,000 customers need service visits and this costs $10M. Thus $90M is saved if the radius of no-problem service increases to just 4.2 miles ( | Top |
EAN: 2147483647
Pages: 154