Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ The 100 Mile-Per-Gallon Carburetor ]   |  Soon, It's Gonna Rain  |   [ Bank Shot ] >>

Weekly Column

Soon, It's Gonna Rain: A New Modulation Technology Promises to Turn Your Cable TV Connection Into a 10 Gigabit-Per-Second Digital Fire Hose

Status: [CLOSED]
By Robert X. Cringely
bob@cringely.com

One of the great problems with technical progress is that it often requires throwing away much of a previous investment. I have boxes filled with old modems, ISDN routers, and Ethernet hubs that are all perfectly functional, but useless to me. I have closets filled with old computers that run like a charm, but do so at 16 MHz. Still, I gladly replaced this equipment each time I could increase performance or decrease cost. And it is this willingness to throw away what's perfectly good in favor of something perfectly better that is the very basis of high-tech industry. It is exactly what makes Microsoft and Intel so powerful, yet both companies have to know they could be replaced in a moment if the right competitor comes to market with the right product. What makes a product right enough to create such disruption? My rule says a 10X increase in price-performance will do the trick. If a new application, operating system, computer, or piece of networking equipment comes along that has 10 times the performance at the same price or the same performance at one tenth the cost, it doesn't matter who makes it, that product will take the market. Brand loyalty is nothing against the power of 10X.

As systems become more complex, it is easy to think of them as being more difficult to replace, but that isn't really true. What's true is the more complex a system, the more points of opportunity there are for improvement. When personal computing was truly personal, and every user's desktop was its own little world, the areas for improvement came down to applications, operating systems, processor speeds, hard disk capacities, graphics subsystem performance, and the speed of your modem. Those all still exist, but now we can add to them the Internet, the servers that sit at the other end of the wire, the wire itself, and a myriad of new services headed our way courtesy of initiatives like Microsoft's .NET. Today, it is possible to gain that 10X increase without changing a thing at your house, or at least by changing very little. And sometimes the things that change aren't at all what you expect. Rainmaker Technologies, for example, is a Silicon Valley startup that simply wants us to use a new technique for modulating the network signal coming into our homes and offices, thereby gaining for each user a 10X increase in bandwidth. Bandwidth is good.

At the heart of Rainmaker's business plan is an awful truth: The part of the network most resistant to change is the wire itself. Plugging in a new box is easy, but replacing the wire that is in our walls and under our streets, well, that takes decades and isn't undertaken lightly. If it was easy, we'd all have optical fiber connections to the Net, yet few of us do. The very heart of network infrastructure, then, is the wire, and Rainmaker claims to have a way to use the same old wire to carry 10 times as much data or more. While their modulation technique can be applied to many media like phone lines, optical fiber, and even wireless applications, Rainmaker's initial plan is to improve the capacity of cable television systems so they can carry more video channels and more data.

As a modulation scheme, Rainmaker's technology exists on the first layer of the OSI seven-layer networking model. If you are not familiar with the OSI model, just understand that layer one is the physical layer that specifies how signals fire over the wire. Level two is the data link layer that tells the electrons whether they are Etherrnet, Token Ring, Appletalk, whatever. The beauty of the OSI model is that it allows designers to make changes in one layer that have little or no effect on the layers above or below. So Rainmaker's modulation scheme, which affects only the physical layer, can run on any medium like twisted pair, coax, optical fiber, or wireless, and can serve up bits for any data link layer like Ethernet, Token Ring, SONet, you name it. The modulation scheme on the cable, then, has no effect on your internal network other than to deliver viruses and worms 10 times faster.

Modulation is the most basic definition of how to inject information onto a wire. The simplest form of modulation is just turning the power on and off, but that doesn't give us very much data capacity, carrying information at only the modulation rate. Each time the power is turned on, it sends one bit down the wire, giving us one bit per hertz, which is not enough for Everquest, believe me. The modulation scheme that Rainmaker has to compete with is called QAM for Quadrature Ampitude Modulation, and is simply a way of using various lower and phase levels to trick an electrical signal into carrying more bits of data per hertz. But eventually, Shannon's Theorem throws QAM up against a wall as line noise limits the amount of data that can be carried.

Rainmaker's patented modulation technique is not immune to Shannon, either, but its noise resistance makes the total bandwidth capacity of the line substantially greater. Rainmaker uses wavelet modulation. Wavelets are mathematical transforms that are typically used at higher levels of the network to compress images or digital video signals. In this instance of modulation, though, we aren't so much interested in wavelets' ability to compress data as much as their immunity to noise and ability to coexist both with other wavelets and with other modulation schemes. Because it is very resistant to noise, wavelet modulation can use the whole data pipe and not have to give up bandwidth on the margin to separate it from other traffic. Indeed, wavelet modulation can be thrown right on top of the current cable TV signal, so old and new systems can coexist. Old customers can have their 30 to 50 channel analog cable service, while new customers — connected to the very same wire — can have hundreds of video channels and very high speed data service.

Where current cable modem users share a data pipe that can carry about 30 megabits-per-second, Rainmaker customers will get 170 megabits-per-second or more. With wavelet modulation filling the entire one gigahertz capacity of coaxial cable at 10 bits-per-hertz, the ultimate capacity of the system is 10 gigabits-per-second for each segmented subnet. That's enough room for 500 HDTV channels on the same cable that's connected to your house right now. If we forget for a moment the heroic effort required to marshal 500 HDTV channels on the cable head end, all that moving to wavelet modulation requires is a new $10 modem chip at each end of the line, according to Rainmaker CEO Paul Chin. Rainmaker, of course, wants to sell those chips, 152 million of which would be needed to retrofit the entire U.S. cable system.

Rainmaker's is a compelling argument for cable operators who can see their infrastructure lasting years longer than they ever hoped. Suddenly, the fight over whether and how to carry HDTV is over at the same time that wavelet modulation enables new services at both the top and bottom of the digital food chain. Ten gigabits-per-second would make possible practical videoconferencing with high quality video at the same time that wavelet modulation's lower power requirements would support lifeline voice-over-IP telephone service even after the power goes out.

I have no connection of any sort to Rainmaker Technologies, which is right now finishing-up the chips that will enable wavelet modulation. But if it gives me 10 times the bandwidth for no increase in cost — heck, I say, "Let it rain!"

Comments from the Tribe

Status: [CLOSED] read all comments (0)