Heterodyne detection, non-linear mixer and minimum detectable signal (2-way radar)

I've been working on a transceiver system with BGT24LTR11 chips at it's core and I was unable to achieve necessary signal to noise ratio. I want to know whether it's theoretically possible and for that I need to understand how the terms in the title all intertwined.

The following parameters/figure of merit I'm familiar with: Minimum detectable signal MDS = -84 dBm, Tangential Signal Sensitivity (MDS with SNR of 12 dB) = -72 dBm, peak radiated power = +30 dBm, receiver antenna = +30 dB

When I do link budget at 100 ft, with 24 GHz carrier frequency, I get the following:
+30 dBm -90 dB (FSL) - 90 dB (FSL) + 30 dB (receiver antenna) = -120 dBm into receiver (1 femto Watt)

Computed TSS/MDS above requires -72 dBm at the very least and I have -120 dBm only. This is where I can't quite connect non-linear mixer to power into receiver. I've refreshed my memory with nonlinear heterodyne detection but how do I relate it to power into the receiver/mixer?

Measurement I've made today only adds to the confusion: I've measured SNR for a single-hit non-fluctuating target (1 m^2 cross section) in dB as Vtarget_rms/Vnoise_rms
SNR: 13.6 dB at 50 ft, 10.8 dB at 100 ft, 7.9 dB at 200 ft

The results are perplexing because radar range equation tells that doubling of distance should decrease SNR by 12 dB, instead I observe 3 dB with each doubling, indicating that I'm not understanding something or there are crucial parameters I'm ignoring.

I realize that there are dozens of parameters I haven't indicated here, ADC, filtering, test conditions, analog circuitry, IF noise spectrum, etc. but the fundamental thing I want to know is if I have enough power to achieve required SNR, so I can start addressing the rest.

There are many things that influence the free-space wave propagation of radar signals that should be taken in account (antenna height, antenna pattern, RCS, distance, etc).

In your SNR measurements did you take in consideration the ground reflection?
If you have strong ground reflection, the SNR may show less variation with distance.

For the measurements, try to use for the target a triangular corner reflector, which gives higher cross section. A corner reflector with edge length of 15cm (11dBsm) or 20cm (16dBsm) will do a good job at 24GHz.

I can make measurement with corner reflector, but this empirical approach won't get me meaningful information. If I could figure out whether I have enough power to achieve the required minimum SNR, I would continue with test. As you rightly pointed out, there are too many variables involved.