G.Lite's test run

In mid-1997, many asymmetrical digital subscriber line vendors and carriers became attracted to the idea of slowing the technology's speed to reduce hardware complexity. Carriers saw this reduction in speed as a way to eliminate the customer premises equipment splitter, and thus the need to dispatch a technician to install ADSL service.

The idea of enabling the end user to install ADSL modems, as they install traditional dial-up modems, gave birth to G.Lite, a version of ADSL that consumers can install. The "lighter" version is designed to enable ADSL to compete more effectively against other high-bandwidth access options.

As the G.Lite standard gathers momentum, a real-world test has shown it is a viable, cost-effective solution for simultaneously handling residential voice calls and high-speed data services such as Internet access. The standard, which was recently endorsed by the International Telecommunication Union, specifies data transmission speeds up to 1.5 Mb/s downstream and 512 kb/s upstream.

The splitterless aspects of G.Lite were tested during a four-month, 47-home field trial in which subscriber telephone lines were used. The trial involved keeping the modem connected to a standard POTS line at all times, as in ADSL applications, so that subscribers would use the same telephone line for data services as well. In most cases, the results of the trial exceeded the expectations of the test team.

Conditions of the trial

The field trial was a collaborative effort of four companies: Fujitsu Network Communications, GTE, Intel and Orckit Communications. Intel employees and their families were the users in the trial.

>From a single central office in Hillsboro, Ore., GTE provided the G.Lite >service to each test home over an existing POTS line.

The homes were located 4200 to 17,500 feet from the CO and encompassed a variety of structural ages, home wiring configurations and telephone devices, which included telephones, fax machines, analog modems, caller ID boxes and home alarm systems (Figure 1). The trial used Fujitsu's Speedport modems and CO multiplexing equipment, developed in conjunction with Orckit.

During the first phase of the trial in September and October 1998, the equipment was installed in 20 subscriber homes in the Portland, Ore., area. These installations were conducted by the test team to facilitate additional data gathering.

An additional 27 subscriber homes were added for the second phase of the trial. In this phase, 10 subscribers installed the equipment themselves. The purpose was to determine the ease in which subscribers successfully installed the equipment on their own, and to record and measure any assistance required. The second phase ran from mid-November 1998 through January 1999.

The overall results of the trial showed that equipment based on G.Lite technology enables maximum data rates of 440 to 512 kb/s upstream and 1.5 Mb/s downstream up to about 15,000 feet from the CO. The upstream data rate begins to decline beyond 15,000 feet, while the downstream rate remains close to the maximum as far as 17,500 feet from the CO - the most distant location tested in this trial.

The right power level

Because G.Lite is a "splitterless" technology, a splitter is not used at the network interface device to separate the voice and data signals traveling into the home. The benefit of installation without a splitter is that the telephone company does not have to send a technician to the subscriber's home, thus eliminating the time and cost associated with installing the splitter for both the telco and the subscriber. With G.Lite, the modem is plugged into a standard telephone jack, which may share the same inside wiring as the subscriber's POTS service. Without a splitter, however, signal interference becomes a concern during both off-hook (when making or receiving a call) and on-hook (when the phone is not in use) operation modes because the electrical behavior of telephone handsets is often undefined at higher frequencies. Table 1 shows common factors that contribute to interference in the home.

The modems used during the trial were based on the T-1/E-1 T1.413 standard, and they included G.Lite features such as profile detection, automatic power cutback per profile and special algorithms to reduce ADSL's sensitivity to POTS interference. The modem configuration included a 4 dB noise margin limit for a bit error rate of 10-7, in accordance with the G.992.2 specification. The power cutback feature was designed to be an automatic function of any commercially deployed G.Lite modem, so the user does not have to perform any adjustments to the modem functionality.

It was necessary to cut back the power level to the modem by 15 dB or more during off-hook conditions to achieve satisfactory POTS performance in most of the homes that did not use microfilters during the trial. However, following an off-hook event, a cutback level this severe rarely leaves enough power for the modem to sustain the G.Lite service, re-establish a connection or resynchronize. The trial also discovered that as the distance from the CO increased, interference from off-hook conditions worsened.

A more effective cutback level was 10 dB or less, which eliminated noise interference in about 60% of the telephone devices tested. Using this cutback level also required the addition of microfilters to telephone devices that still experienced off-hook noise interference.

The role of the microfilter

Microfilters are simple, in-line devices that can reduce signal interference in G.Lite installations, where the same telephone line is used for voice and data services. These easy-to-install devices block ADSL data signals from interfering with that device while allowing POTS voice signals to pass through. A low-pass, unidirectional microfilter manufactured by Orckit was used for the trial.

In 81% of the homes tested, acceptable off-hook performance could only be sustained when at least one microfilter was installed in the home. The trial defined acceptable performance as the modem's ability to resync after an off-hook event, and to achieve ADSL bit rates while telephones were off-hook. Multiple microfilters had to be installed in half of the homes to reduce interference caused by more than one telephone device (Figure 2). Where required, the microfilters enabled minimum standard off-hook data rates of 256 kb/s downstream and 32 kb/s upstream. This included no more than five errored seconds in a two-minute test period and no perceptible noise on the house telephones.

Only 23% of the homes experienced problems in which ADSL service performance declined when the telephone devices were on-hook. Microfilters were installed in these homes to improve the on-hook data rates and reduce errors.

The field trial also considered the microfilter requirements of different types of telephones, including corded and cordless. Many corded telephones required a microfilter to improve off-hook performance by eliminating noise interference, but most cordless telephones did not. For cordless telephones, microfilters were more often necessary to improve on-hook performance.

Microfilters were installed inside the home between a telephone device and a jack. The devices can be installed on several telephone devices and should not affect the quality of POTS. At the end of the trial, two users reported that the performance of a few instruments seemed to be affected by the microfilters. This included one instrument with increased feedback between the transmitter and the receiver in the same handset with the microfilter in place, and one instrument that transmitted poorly to the distant end user after the microfilter was installed. Users made these comments during post-trial interviews, and no specific data was available to isolate the cause.

Positive results

The G.Lite trial in Portland uncovered two important findings:

- Power cutback at the modem was not sufficient in most cases to accommodate off-hook conditions.

- The installation of one or more microfilters on telephone devices in the home was required in most of the G.Lite installations.

During the second phase of the trial, when subscribers installed the equipment themselves, nine out of 10 installations were completed successfully without requiring on-site assistance at the home. In fact, seven of the 10 subscribers were capable of completing the installation and achieving data rates to their satisfaction without any technical assistance (Table 2). This included installing the modem and determining the need for and installing microfilters on affected instruments. The key was delivering the microfilters and easy-to-follow installation instructions along with the modems.

In a few homes, the subscriber needed to call a technical support person from the test team to get the modem working without errors. This scenario represented around 20% of the installations. Situations in which such assistance was helpful included putting a microfilter on a fax machine and turning off the power to a non-working home alarm system.

In almost every case, whether the subscriber completed the installation alone or with telephone assistance, the installation could be successfully completed without undue complications. Exceptions included installations involving wiring for home alarm systems that would not accommodate installation of a microfilter, hard-wired telephones requiring filtering (or removal), and two telephone lines using different wire pairs on the same in-house wire cable. These difficulties could require technical assistance at the home or in a commercial deployment. When a truck roll was necessary, the trial team tackled more difficult tasks such as replacing a network interface device or finding an electrical short.

One important issue the trial addressed was that of performance associated with specific placement of the G.Lite modem in the home. During the field trial, ADSL modems were successfully used at two locations within each home - specifically at the network interface device with home wiring disconnected (to remove all influence of the home wiring from the result) and with all wiring reconnected at another jack in the home that was farthest from the network interface device (to include any influence introduced by the wiring in the home). At both locations, the interference levels and maximum possible bit rates were the same in nearly every case.

Because of the variations in the wiring encountered, the test team was able to conclude that inside wiring would not be a major contributor to any difficulties with G.Lite. Home wiring alone showed little effect on standard G.Lite rates, although it produced a minimal effect on rates above the standard. Only when telephone devices - primarily telephone handsets - were added to the wiring was the rate significantly affected.

By eliminating the need for a splitter outside the home, G.Lite technology brings significant time and cost savings to the telco, as well as the subscriber. Subscribers can install a microfilter in the home between the telephone device and jack as required to ensure maximum data rates without interference.

In all, the four-month field trial showed that G.Lite is capable of fulfilling its promise of successfully bringing consumer-installable, high-speed data services to the home over the same telephone line that currently provides voice service. This makes G.Lite an ideal solution for today's residential needs.