Stardate 20020927.1115

(On Screen): Joe Katzman discovered an article which made him afraid that our GPS-guided weapons could be defeated with a simple commercial jammer produced in Russia which can be bought for $39.95. I'm not too concerned.

A normal civilian GPS receiver uses an omnidirectional whip antenna, so that any signal from any direction is received equally. I would be very surprised if JDAMs do so. I think it's pretty obvious that any jamming would be coming from near the ground, and that the true GPS signal would be coming from the sky, and it isn't very difficult to design an antenna which selects signals from only one direction or set of directions, and physically rejects signals from all others. Since the bomb that a JDAM is attached to will always assume a standard orientation when it's falling (nose down) it would be quite easy to incorporate a directional antenna into the JDAM, and that alone would go a long way towards solving the problem.

CPO Sparkey works on this stuff and thus can't comment, except to say, "Come on Saddam, just stand still with all the dinky little Internet jammers you can buy just piled around you, please, oh please, oh please!”. It probably should be pointed out that a jammer is a prime target for HARM or equivalent weapons, and by its nature it cannot hide. If a high value target surrounded itself with jammers like this, it would be like painting a target on the site.

I don't on military hardware, but I've worked on related civilian technologies, so perhaps I can offer some speculation.

The original GPS system uses a form of direct-sequence spread spectrum (DSS) for its signals. The CDMA cell phones I used to work on use a much more sophisticated version of the same thing. What has to be understood is that DSS is designed to ignore substantial amounts of environmental noise. The signal in DSS is scrambled using what's known as "pseudo-noise", or PN, which is a carefully crafted sequence of bits known to both transmitter and receiver, and because of how the receiver works, it will very strongly reject anything which is not encoded with exactly the right PN sequence. (Or even anything which is encoded with that PN sequence but at a different phase.) The main reason our design in CDMA was more sophisticated than the original GPS system was that CDMA was designed later and the state of the art had improved. What we were doing required hardware in the receiver which wasn't technically possible when GPS was originally designed. This is important for CDMA because each phone operating on a given frequency looks like a jammer to all the others, but each phone uses a different PN sequence, so the ability to ignore any PN sequence except the one intended for you makes it so that your call will still get through to you. (I wrote an explanation of the basic concepts involved here.)

You'll note that I kept saying "original GPS system". Here we enter the realm of speculation: satellites have a limited lifespan, and to keep the GPS constellation going, new ones have to be launched periodically to replace older ones which have exhausted their fuel. All satellites have rocket engines on them, used to adjust their orbits which will drift because the earth isn't a regular geometric shape, and in the case of GPS the orbits of the satellites are particularly critical, because the system as a whole is as accurate as the satellite orbits are. The Navy has been sending new satellites into orbit on a regular basis ever since the system was first created, nearly 30 years ago. The GPS system is old, but all the satellites up there are very new. The basic technology which made it possible for CDMA to use more sophisticated coding has been available for 15 years, and all the satellites up there were launched much more recently than that. (I suspect they have a lifespan of about 5 years, but it could be less.)

The new satellites still generate the original GPS signal because it's become important for civilian use, but has the NAVY, all in secret, actually implemented a newer parallel system run from the same satellites which uses much more modern and far more sophisticated forms of DSS, on an entirely different carrier frequency? I wouldn't be at all surprised, but if they have that fact is probably classified. But in fact I'd be really surprised to learn that they had not done such a thing.

One of the reasons why this kind of improvement in DSS technology is interesting is that it increases what we call "coding gain", which means that far more signal gets through without the transmitter having to raise its power level.

The problem for wide-band jamming is that it's putting out fuzz. It's sort of like a very loud hiss in radio frequencies. It's random, but it's not the right "random"; it doesn't match the PN sequence, so the DSS receiver can reject it pretty easily. (For reasons which are kind of difficult to explain, it cancels itself out in the rake receiver.) In order for it to actually fully block the signal, it has to actually be sufficiently loud so that it raises the noise floor high enough to drop usable signal (which, in CDMA, we referred to as EC/I0, pronounced ee-see-over-eye-not) below some critical threshold, and that's a lot harder than you think. We could successfully receive signal with 18 db of EC/I0, and that's not very much. Indeed, because of how DSS works, it isn't enough just for the jammer to produce a stronger received signal at a given point than the one it's trying to block. It's not that straightforward. You can receive signal successfully when the rejected signal is 20 times as loud. (In fact, in CDMA the real signal is always lower than the crap being rejected, which is the sum of the traffic for all other phones in the sector. If there are 30 phones in calls in a sector, your signal is only 1/30th of the total, or putting it another way, there's 29 times as much crap as signal.)

I do not believe that a 4 watt omnidirectional jammer actually could block a modern DSS system in a hundred mile radius. CDMA cell towers are typically placed on about 4 mile steps and they broadcast 30 watts or so (per sector, usually with three sectors per mast), all on the same frequency, without interfering with one another, not to mention the fact that each of those 30 watt carriers is carrying a lot of calls simultaneously. (That's one of the reasons why CDMA is so cool.) A 4 watt jammer might be good for one mile, and maybe not even that far. The only reason that a 4 watt jammer might have a hundred mile range (discounting typical advertising boasting) versus GPS is because the original GPS system used a pretty unsophisticated form of DSS and didn't get a lot of coding gain from it. If there's a parallel modern one, then to get hundred mile jamming you'd need to jam with kilowatt, if not megawatt, power levels.

If such a system exists, and if it's using a very broad carrier (20-50 MHz), with an equivalently fast PN spreading sequence, and hundreds of thousands of chips per bit (compared to the 128 chips per bit that CDMA uses), then it's going to be getting ungodly amounts of coding gain and it's going to be damned hard to jam even if it wasn't using directional receivers. If such a system exists and if it was designed properly, then it could not be jammed over a wide area for any kind of practical purposes. You're not going to be deploying huge numbers of kilowatt jammers, and even if you did they would be trivial to find and destroy.

Donald Sensing points out that JDAM has an inertial guidance system as well as a GPS guidance system, and that's another good point. Most of our other PGMs have redundant systems as well. The Tomahawk cruise missile now uses GPS, but it was originally designed to use an extremely sophisticated form of internal guidance which was based on using radar to look at the shape of the ground in order to determine if it was on the right flight path. That system is still in there, and if GPS becomes useless, no matter why, the Tomahawks will still fly to target; they'll just become slightly less accurate.

I'd like to observe that the most likely reason that an inertial guidance system was incorporated wasn't because they were worried about jammers. The Soviet Union had developed a credible anti-satellite capability, and in the event of a major shooting war there was a concern that the USSR might take out a large part of the GPS constellation.

Donald puts it well: American designers are not idiots. Jamming is not a new concept, and it was an obvious problem when the idea of using GPS this way was originally proposed. If they couldn't solve it, they wouldn't have implemented these systems at all. Sparkey knows what they did but can't tell us. I don't know but I'm sufficiently confident in my engineering brethren to believe that they did.

Update: Tom Rothamel writes:

Is it hard to jam a spread spectrum signal if the spreading code is known?

It doesn't actually help you a great deal, if what you're thinking is that you can lay your signal on top of the legitimate one and disrupt it that way. The problem is that your transmitter isn't in the same physical location as the one you're trying to disrupt, and the speed of light isn't infinite. So your signal won't be synchronized with the other one in most locations, and when they're out of phase then the fact that they're using the same PN sequence no longer matters.

In CDMA, it would look like a separate energy spike to the rake receiver if it was within a certain phase window, and if it were high enough it might louse up the summing in the rake receiver, but in CDMA every individual phone in a call anywhere in the entire system has its own unique PN pattern, created by XORing the long code, some relative phase of the short code (different for each cell), and the Walsh code assigned to that particular phone for that particular call (which represent channels for that cell).

For a broadcast DSS system like GPS, it would depend enormously on how resilient the search algorithm was. What I'd expect is that if the receiver found an energy spike which turned out to not make sense, it would start hunting again and look for others at different phases. It has to do that anyway because it has to find three different satellites to make a fix, and it's the phase difference between those satellites which gives it the detailed timing information needed to determine where it is. So if the receiver is designed well, I wouldn't expect this to be helpful. Your broadcast would just look to the receiver like a faulty satellite, to be ignored, but it wouldn't interfere with the ability of that receiver to get true signal from real satellites..

The thing is that the PN pattern is designed so that it doesn't interfere with itself when it is at a different phase.

I'm sorry for all the jargon, but there isn't any other way to explain it briefly.

Tom continues:

I don't know enough about CDMA to know if there's a central control channel that could be jammed in the same way.

There is, in a sense. One of the Walsh Codes is reserved for what's known as the "paging channel", and if you disrupt that for any given sector, then the cell system can't set up calls, even if the traffic channels still work. But the ultimate spreading pattern for each sector for the paging channel is different because the short code phase assigned to that sector is different. And it's still not the case that when you do this you're crunching the cell's own traffic channel; rather, you're providing a big nonsensical energy spike in the right phase space and trying to overload the rake receiver sum with nonsense.