Finding The Holy Grail Of Modem Connectivity With V.34, the line was drawn in the sand by standardizing on measuring a maximum of 3,200 events per second (3,200 baud) and having each event represent 9 bits of data via a combination of modulating the amplitude and the phase of the signal. The math shows that 9 x 3,200 = 28,800 bps. The problem lies not in modulating these different amplitudes and phases on to the transmitting wires, but in demodulating them at the other end of the wire. All forms of interference on a pair of copper wires get in the wa y of determining what's what. You can imagine that to generate a waveform shifted by 60 degrees or 65 degrees is fairly easy, but once it's been transmitted over a long piece of copper wire, it's difficult to determine that fine degrees of phase change. Searching for the Grail This is the crux of the matter regarding the unidirectional nature o f 56-Kbps technology. By connecting a digital channelized T1 line straight into a remote-access server, the server is connected digitally to the telephone company's central office switch. This is important because by connecting to the central office switch, one of the analog to digital and digital to analog conversions has been eliminated at the access server end of the connection. Most of a carrier's network is digital--the local loop (to a residential home) is the only analog part. So, by connecting a remote-access server digitally to the telephone company's central office switch, the signal from the access server is digital all the way to the local loop feeding the residential home. It is possible to convert a digital signal to an analog waveform with a higher degree of precision, than it is to convert an analog waveform that has been digitized. Therefore, we can achieve a finer degree of phase and amplitude modulation from the remote-access server to the remote user than we can in the return direction. The rule is that to achieve 56-Kbps connectivity,only one digital to analog conversion can occur on the path between the remote-access server and the modem. The implication is that you must order a trunk side T1 rather than line side T1 from the telephone company for connection to a remote-access server. With a line side channelized T1, an additional analog to digital conversion typically occurs, making 56-Kbps connections impossible. Of course, many users do not get a 28.8-Kbps connection today because of low-quality local loops, and many users will not get full 56-Kbps connection in the future for the very same reason. However, connection rate s for the access server to the user at speeds above 40 Kbps can soon be expected. Who's Cup is It Anyway? There are basically two camps in the 56-Kbps modem marketplace at the moment--one occupied by U.S. Robotics and the other by the Rockwell/Lucent alliance. U.S. Robotics is the only vendor with any real world mileage for its 56-Kbps technology (cal led X2). The main advantage U.S. Robotics has over its competition is that it can download software to Flash ROM chips to upgrade existing modems to 56 Kbps. Other vendors have used specialized modem chips that either require a new modem or a chip swap-out to upgrade to 56 Kbps. Approximately three quarters of all modem manufacturers have followed the latter model, which is based on the Rockwell chipset. For 56 Kbps, Rockwell had a solution called K56Plus, Lucent had V.Flex2 and the pair teamed to offer K56Flex. Having recognized that Flash ROM upgrades are a bet ter idea than physical chip replacements, Rockwell and Lucent plan to produce products using Flash ROM rather than nonupgradable chips for 56-Kbps modem deployment.

Multihoming Your Internet Service With BGP by Chris Lewis and Thanh Nguyen Updated July 10, 1997