Undetectable Radar? (Probably Not)

Active radar signals, due to those pesky laws of physics, are generally easy to detect. Because a radar system emits a powerful beam of electromagnetic radiation, traditionally in a very narrow frequency band, an adversary equipped with only a passive radiation detector can easily zero in on the platform carrying the radar.
For decades the military has been searching for a less visible (and vulnerable) "low probability of intercept" (LPI) radar. This June, Ohio State Universitys ElectroScience Laboratory claimed that its engineersled by Dr. Eric K. Waltonhad succeeded and "invented a radar system that is virtually undetectable."
A flurry of fawning press coverage followed. Even Dr. Walton, though, acknowledges that he did not invent noise radar, as the technology is calledit was first proposed in the 1950s. He did, however, receive the first patent for the technology earlier this year. Heavy signal-processing requirements kept noise radars in the lab for decades, but they have finally proved feasible (and, according to Walton, cheaphe claims around $100 per unit).
And they probably are undetectableby typical radar detectors.
Typical radar signals are high-power, narrowly focused pulses;* each signal is extremely short. Most radars cant send and receive at the same time, so immediately after a pulse is sent out the radar switches to listening mode and strains to hear the pulsess echo. Incidentally, this makes them farsighted-they cant see objects up close.
To detect these radar signals, an adversary can simply sweep his field of view searching for high-powered pulses that are narrowly focused at a single frequency. Since radar signals cannot be perfectly focused and are not constrained like lasersthe beams become larger as they travel, to form a conethis is easier than it might sound.
Engineers have developed new techniques to make detection more difficult. For example, frequency-hopping radars move each chirp to a different frequency (the F-22 radar system reportedly does this), while spread-spectrum (radars and radios) use a (small) band of frequencies simultaneously. The signals are still extremely powerful compared to background noise, though, and are relatively easy to find with the simple detectors mentioned above.
Noise radar is different in two main ways. Like spread-spectrum radar, it spreads its signal over a band of frequencies, but the band is about 1,000 times wider than most spread-spectrum technologies. Furthermore, the signal is also shaped to look like noisethe radio equivalent of ants racing on a TV screen.
The wide band of frequencies has several advantages. Different frequencies interact with different materials in different waysbasically, using an ultra-wideband (UWB) signal allows you to see through walls, trees, rock, and many other obstacles if the signal is well constructed.
More relevant to this discussion, UWB noise radar signals also spread their power out over the different frequencies; the result is that traditional detectors, searching for very powerful signals near a particular frequency wont see noise radar. They will just "hear" more static.
And since the noise radar signal is shaped like, well, noise, it would also be hardif not impossibleto find it by looking for a pattern in the chaos. The noise radar can only detect its own returned signal by first recording it, then comparing a time-delayed version of the recording to what it hears reflected back. (This characteristic also means noise radars detect in "rings" -- the simplest version would detect movement only at a fixed radius from the radar, but it is possible to scan many "rings" very quickly for a more complete picture. The computing requirements for this type of scanning make placing noise radars on fast-moving platforms impractical for now, but they would make exceptionally good proximity detectors, for example.)
Because of their UWB signals, noise radars work best by looking for specific targets -- they must incorporate some knowledge of what a specific targets reflection will look like. They would have great difficulty detecting an unforeseen obstaclewithout prior knowledge of what its reflection would look like, the noise radar would probably just see right through it.
The best way for an adversary to detect a noise radar would be to search, directionally, for sources of UWB noise. The key question here is how "loud" the radars noise would be, compared to background sources like the sun, the galactic center, local power lines, battlefield electronics, etc. Noise radars could be constructed in any number of different ways, and the signal could also be endlessly changed for different applications; lacking specific data, it is hard to speculate on how difficult they will be to detect with this technique.
From what we know now, the "undetectable" claim is something of a stretch, but these radars will almost certainly find uses. They do not interfere with each other or nearby electronics (which are designed to filter out noise), and they can see through walls. If ever used in a military capacity, they would likely force a change in radar detection and seeking technologies. It might cost the Pentagon a pretty penny to detect these new toys, but undetectable radars are probably still a long way off.
-- Eric Hundman
*UPDATE: Thanks to Rutty for the clarification. I originally wrote "chirps" here rather than pulses, which was incorrect. "Chirping" in this context refers to a popular type of signal modulation often used in radars--it ultimately allows for greater resolution.