In this case, I'm sure you're right.

But it might be worth noting that you can use spread-spectrum techniques (http://en.wikipedia.org/wiki/Spread_spectrum) to overcome jamming. Spread spectrum has some similarities to an encryption scheme, because the method used to smear the signal around uses a pseudorandom sequence which is known to the encoder and the decoder.

CDMA (http://en.wikipedia.org/wiki/Code_division_multiple_access) cell phones use this idea.

As you note, this type of encryption takes place at the RF level.

Surely the "jammers" could just broadcast their jamming signal over a large range of frequencies?

Of course, but that raises the amount of resources necessary for jamming all that much higher, perhaps taking it out of the realm of feasibility.

Right, and you now have variables under your control to drive their power requirements arbitrarily high -- namely, the better your timing, the higher your effective S/N ratio (because your spread-spectrum signal can be, in effect, very-very narrowband, but hopping all over).

You increase your timing accuracy by 10x, and the jammer's power requirements increase by 10x (crudely speaking -- not a 1:1 trade, but directly linked).

That's a different kind of spread-spectrum. That's a frequency-hopper, where the carrier changes frequency (the first SS- an idea attributed to Hedy Lemarr in the lore).

We're talking about direct sequence, which increases the bandwidth of the signal but keeps the same carrier. To jam that (theoretically) you need large power through the entire band.

I was talking about both -- for the purposes of explanation, frequency-hopping is more intuitive. (I was trying to be responsive to @Redfern above, who mentions frequency-domain jamming.)

Both frequency-hopping and direct-sequence spread energy over a larger bandwidth. Direct-sequence does not explicitly frequency-hop, but the time-domain switching it uses will spread the energy out in Fourier space.

This kind does:

http://en.wikipedia.org/wiki/Direct-sequence_spread_spectrum

DS Spread Spectrum spreads the signal over a larger frequency range- because power is conserved it can even hide behind ambient noise.

When un-spread, the noise is spread out and the signal is "collapsed" back to the original bandwidth.

I wonder if it still resistent to jamming when someone does so on each used DSSS channel...

It take a huge amount of power to cover the whole spectrum at jamming levels. That's the beauty of the technique.

Jamming can happen at all layers.

The most basic of jamming is to drown out all other communications with high-power noise. But in the case of illegal/covert activities this isn't ideal as your transmitter will be pretty easy to find.

So then you have to pretend to look like a legitimate signal. Which is where encryption and modulation can help. But then application level encryption isn't going to help against basic jamming.