sure, it is all possible. If you made a radar receiver with TWO Local oscillators, say with a 1st IF at 30 ghz, and a 2nd IF at say 2 GHz, the LO frequencies would be way off from where someone would go looking for it.
You could get creative, and have a microwave/optical mixing device,, where the LO frequency is only from pulsing light, which will not leak out a microwave antenna.
You could add high quality (probably) microwave waveguide bandpass filters that would drop the LO leakage by 80 dB. that with some backward isolation in a front end LNA might get you to the -110 dBm sort of leakage level, which would be hard to detect.
You could use a subharmonic mixer, where the LO frequency is 1/2 of where it needs to be. the mixer has diodes set up to effectively double that frequency when being used as a downconverting mixer. So it there is any LO leakage, it is at 1/2 the expected frequency. also would be easy to filter at the front end.