The HDSL2 chipset from Excess Bandwidth arrives during a time of serious soul searching for DSL providers as a whole. As they try to meet their last-mile requirements, all current options are up for review under five main criteria: data versus reach, symmetry, latency, spectral compatibility, and the twisted pairs required.
Data rate versus reach is a major bone of contention when it comes to DSL in general. This tradeoff is critical to DSL providers because it defines the coverage area for their DSL offering. As for symmetry, some DSLs provide highly asymmetric connections in order to optimize for specific applications, which isn't conducive to multiple-voice support. Voice also is closely tied to latency. ADSL, for example, has significant latency, due to the frame structure chosen for the standard. Though not noticeable during web browsing, the excessive latency can cause problems with telephony and video conferencing.
The spectral compatibility issue is particularly thorny, especially as DSL deployment reaches critical mass. Twisted-pair loops are run from the central office to the subscriber in groups of 25 or more pairs. These are known as binder groups. It's critical that if one subscriber in the binder group has HDSL digital service, while another has ADSL, the two must not interfere with one another.
As for the number of pairs required, the ideal is one, to reduce the implementation cost, especially in the home. An optional, multiple-pair solution can then increase bandwidth according to subscriber requirements.
By far, the most common form of DSL being installed today is ADSL, especially in homes. In addition to the asymmetric constraints that limit the application of ADSL, its early deployment has not been smooth. Issues of reach, crosstalk, and poor performance often plague installations. The future holds the promise of even more problems. DSL deployment is growing and the number of twisted pairs within wire bundles carrying DSL are increasing traffic, resulting in even greater crosstalk.
To accommodate emerging needs, providers asked ANSI to come up with a robust, symmetric DSL service that can reach the overwhelming majority of the customer base—with guaranteed performance. Such a standard was defined late last year as high-bit-rate DSL2 (HDSL2). Its cousin, G.shdsl, is a more generalized version being developed under the auspices of the ITU. G.shdsl is a 2.3-Mbit/s to 192-kbit/s, rate-adaptive service, primarily aimed at the consumer and small-office/home-office (SOHO) markets.
The G.shdsl/HDSL2 standard provides for the transmission of data at a distance of over 20,000 feet at high data rates on a single pair. The ability to provide both data and multiple voice lines over just a single twisted pair (versus two pairs for HDSL) is crucial for service providers. It will allow them to establish both voice and data revenue streams, creating an optimum economic model.
DSL not only promises high-speed data, but also the provisioning of data and multiple voice lines. While single-pair ADSL offers one baseband voice channel occupying the part of the frequency spectrum below the data, the symmetric bandwidth and low latency of G.shdsl/HDSL2 are able to provide multiple digital voice channels within the datastream. Additionally, for the business environment and such emerging consumer applications as video conferencing, the "universal" communication model of the symmetric G.shdsl/HDSL2 technology is a necessity.
The robustness issues addressed by HDSL2 were the wide range of loop configurations and crosstalk. The loop conditions can vary widely, with splices and bridge taps that have different lengths of varying American wire gauge (AWG) sizes, causing both frequency and echo variations. Test loops similar to the carrier-service-area (CSA) test loops used in the spec are shown in the figure.
As it propagates down a twisted pair that becomes part of a wire bundle and a binder group, the crosstalk that a signal encounters is a time-variant problem. It changes when services on adjacent pairs change. The standard includes tests in which the neighboring copper twisted pairs in a bundle are crosstalking with the pair being tested.
The HDSL2 standard takes a set of loops, similar to the ones shown, and defines reach performance for those loops. In addition, for the two specific test loops shown, the standard defines a scenario in which the crosstalk affecting those loops is 1% worst case—that is, 99% of the crosstalk scenarios will be more benign than this example. This scenario is called the minimal test set, and defines the most stringent of the HDSL2 test cases.Studies have been performed on the maximum attainable margin requirements for this worst-case situation, per the physical limits of the medium, as laid down by communications signal theory. One of the most notable conclusions was that in some cases, the maximum theoretical performance is only 2 dB greater than the required 3.75-dB standard. Therein lies the biggest difficulty in meeting G.shdsl/HDSL2: the communications system must perform to within 2 dB of the theoretical limits of the medium.
Thanks to the severity of the HDSL2 specification, having an advanced chipset, such as Excess Bandwidth's, is only half the battle. For a design to really come to fruition, it is imperative that a holistic approach be taken, whereby the signal path is analyzed—from the system input to system output. a complete system analysis of all signal paths.