It's been a long time since I was involved with Army radar and encrypted communications (in my case, merely humble air traffic control equipment), but the article intrigued me enough to do a very quick lookup. This article isn't very technical, but I can see how it's not simple spread-spectrum radio.
Remember that the information conveyed by radar microwaves is limited; we're primarily interested in reflections (this is "primary radar"; "secondary" radar actually does transmit information; IFF is a type of secondary radar). For a simple radar we know the radar echoes are ours because they come back to our own dish, and match the frequency that we transmitted. They're also incredibly easy to jam.
Frequency hopping on its own makes things harder to jam because the frequencies change in a cryptographic pattern. They can still be jammed if your broadcast a lot of noise over the entire spectrum, but then you limit your own communications. If you can detect the point source, though, you can broadcast a point source over the entire spectrum and still jam them.
What I think I understand about this is that it’s not merely frequency hopping, but the signal modulation is encrypted in a way to evade detection. With a receiver I can detect a typical radar’s 3.4 Ghz signal at -200db (numbers are made up), even if spread across the spectrum, because I know what a 3.4 GHz square wave looks like against the background noise, even if it only appears intermittently on the narrow frequency I’m scanning.
I could try to modulate the signal a different way; maybe a sawtooth, maybe a sine, but a repeating, predictable signal is observable, even with frequency hopping. However if I broadcast noise (and my receiver knows the noise’ pattern), then any listening equipment shouldn’t be able to pick out my microwave pattern from the background.