First, some basics about modems. Modems take computer data and transmits it over telephone lines. Since phone lines transmit audio signals, the modem MOulates the computer data into sound. The other end of the connection will DEModulate the sound into computer data. Thus the name "modem." There are obviously a number of different ways modems can modulate and demodulate, so the modems at each end must use similar methods for sucessful communication to occur. The organization entrusted with making decisions on these standards is called the ITU (International Telecommunictions Union). New standards, such as those being proposed by both US Robotics and Rockwell, get submitted to the ITU in hopes they will become a "standard" that any modem manufacturer can use.
Modems have been around for at least as long as the Internet, though they have only become a consumer item since the first pre-built personal computers in the late 70's. Back then, modems ran a paltry 300bps (bits per second), which translates into about 30 letters per second. In the late 1980's, 2400bps modems were more common with USRs using their proprietary HST-protocol which did an amazing 9600bps. 14.4k modems started making their way into consumer markets in 1992. 1994 was the year of the first 28.8k modems.
As fast as modems have advanced in the past sevearl years, we have reached a limit to the amount of data that can be passed thru a standard phone line. Shannon's Law is used to determine how much data that can pass thru a given communication medium. According to this law, the maximum amount of raw data that can be pushed thru a standard analog phone line tops out at about 35k-bps -- very close to what today's 33.6k modems can push thru the phone lines.
If we are near or at our limits right now, how can they make modems that supposedly go faster? The answer: Change how you look at the phone lines. Today's modems assume a totally analog communications channel. This isn't necessary true when you call into an ISP with hundreds of phone lines. For companies with a large number of telephone lines, the telephone company will often combine them into one big, digital telephone line that connects directly from the ISP into a local telco switching office. Because the ISP's end is digital, their end of the connection can be viewed digitally, which subjects the connection to higher theoretical limits under Shannon's Law. The end result: In one direction (most likely from your ISP to you), the connection will be 56k. In the other direction (you communicating back to your ISP), the connection will still be 33.6k because that end of the connection will still be analog. In other words: things will come into your computer faster, but they won't leave your computer any faster than they do now.
US Robotics and Rockwell are proposing different implementations of this idea. The two will likely be incompatible with each other as was the case with the V.FC and V.34 standards back in 1994. It is also likely that one or both of these implementations will make it to market before the ITU ratifies a new standard. Back in 1994, the ITU ratified V.34, though V.FC modems were on the market before V.34 was ratified. It had gained enough market share to become a "de-facto" standard. I can see something similar happening with these competing 56k standards.
Who will really benefit from these new modems? ISPs (usually large ones) who get their phone lines in big, digital blocks and the users of such ISPs that happen to have "good" lines. Users of smaller ISPs, those with analog switching equipment, and other users that talk "modem to modem" will most likely not see any additional performance out of this new breed of modems.
If you want to find out more about the new standards being proposed by Rockwell and US Robotics, check out Rockwell's pages on their K56Plus Technology as well as US Robotics X2 Home Page. Both include white papers on the technologies they will use to implement 56k modems. I found the white paper on Rockwell's site more interesting and detailed whereas US Robotics' seemed a little "light" to me.
Since I originally wrote this article, a few interesting things have happened: