For a GNSS-type system to work, you need four transmitters in sight. This would imply a moderately dense ground network on the moon for a receiver at orbital height -- and as this article points out, we're not doing great at building even a spare ground network of right-side-up transmitters right now -- but would need increasing density for a descending receiver. We don't really have challenges right now with accurately determining orbital parameters of probes above the moon, it's the final few miles where GNSS-level accuracy would be tempting; but that's also the domain where seeing four ground-based transmitters is basically impossible.

>nothing is on the moon

It's not exactly nothing, the moon itself with its mountains is there and blocking the radio signal and it means propagation isn't unlike that of VHF/UHF on Earth. You still only have no visibility beyond the horizon. In order for it to be visible and useful during landing you would need the base station to be reasonably close. I guess that would make it more akin to ILS/VOR/DME than GPS. That obviously wouldn't be feasible until we have a permanent base there (perhaps an unmanned one).