Atomic clocks are not precise enough, they need nuclear clocks, which are not developed yet: https://thoriumclock.eu/

Is it really not possible to converge a location from implicit signals like frequency shift from relative motion? My intuition, maybe wrong, is that with enough compute and a thorough enough ephemeris of what should be broadcasting what, you could derive location.