Stardate 20030523.1635

(Captain's log): In response to my post here about a month ago about the digital warrior (edited and reprinted in shorter form by WSJ Online last Monday), a couple of people wrote to ask me whether this increasing reliance on digital communications by our military forces might actually open up new vulnerabilities once enemies learned how we were doing it.

All advances in military capability also create new kinds of vulnerability. I used the American Civil War as an example of the ways in which the transition to industrial war changed things. In the Civil War, there was heavy reliance on the railroads and the electric telegraph, and because of that both became targets. Both sides in the war had large forces of cavalry. The primary job of the cavalry was scouting, screening and raiding. It was common for both sides in the war to send cavalry units of various kinds out into the rear of the enemy's army. (J.E.B. Stuart became rather famous for this kind of thing.) One of the things the cavalry would do if it had an opportunity was to try to ruin railroad track and to destroy locomotives and rolling stock, or to try to take down telegraph wires.

Even more valuable was to send very small units out to tap into and listen to the messages on the telegraph, which is why both sides eventually started using encryption for those messages.

It's a problem, but not necessarily an insurmountable one. Some vulnerabilities can be anticipated; some will catch you by surprise and you'll have to put up with them until they can be fixed.

In the case of the anticipated near-term digital battlefield, where every vehicle and every dismounted rifleman is actually part of a digital network, there were a number of concerns expressed by readers. But in fact, they can be anticipated and a lot of them can be dealt with using good design.

I sat down this afternoon and tried to think through a number of potential vulnerabilities in the system, to decide whether it was possible to protect against them. Basically, I think it can be for the most part.

For instance, one concern was triangulation. If all the individual soldiers in a digital platoon are able to track each other's positions on their helmet-mounted displays, it's because they're all transmitting their positions constantly. But that means they're emitting RF, and it would seem as if their positions could then be tracked by the enemy using more prosaic kinds of triangulation.

There's also the concern that a system like this might be vulnerable to jamming. Or the enemy might snoop on the network and gain the same information from it as the individual soldiers, or even be able to feed bogus information into it and fool those soldiers.

Actually, all of those can be prevented using the right kind of transmission technology and communications protocol design. The basic kind of technology used in CDMA cell phones (such as the IS-95 phones I used to work on) can be enhanced in ways which massively help this.

Without going into too great of detail, what such a system does is to broadcast a relatively small amount of information using a very broad spectrum, in a way which massively increases the redundancy. As a side effect of the design, this means that multiple units can operate simultaneously on the same frequency band without confusion. (More on how it's done here.) The main way this is done is by breaking all the bits into "chips" (sort of pieces-of-bits), and sending a lot of chips per bit. If the majority of the chips are received properly, the receiving unit can reproduce the bit.

Such a system exhibits what's known as "coding gain"; it means that information can be transmitted reliably with far less raw transmit power. (More or less, you transmit each bit longer at low power instead of shorter at high power, though the explanation is actually more complicated than that.) The more coding gain the system has, the less raw power it needs to use.

These systems are inherently able to get their signal through even in the face of huge amounts of noise and interference. (Indeed, that's the normal operating mode. The signal transmitted by each unit on a given carrier looks like noise to all the others. It is routine for there to be 20 times as much noise as signal.)

With the EVRC codec, IS-95 transmits 128 chips per bit on a 1.25 MHz carrier. A military system could use a 20 MHz carrier and some huge chip-per-bit rate, and by so doing get a ferociously high coding gain.

That means that transmit power would be very low in absolute terms, and would be spread over a very wide amount of spectrum. It turns out that it's damned difficult to even detect such a system in operation unless you know a hell of a lot about it (information which would only be accessible through espionage). And classical triangulation isn't really possible if there are several units on the same carrier operating at once; it isn't possible using raw radio technology to separate their effects.

Such a system is also damned near impossible to jam. It isn't enough just to pour out more crap, at a higher power, than the signal. These systems routinely have far less signal than noise even when they're not being challenged. As the coding gain increases, the amount of power needed to fully jam it also rises massively. As a practical matter, j