minimum usable 5G cellphone signal level?

I'm a programmer (not and EE) working on a project to simulate coverage and capacity for 5G cellphone signals at 26 GHz. But I've done enough path loss calculations for terrestrial and satcom radio to question what the minimum received signal level for 5G cellphone operations will be.

I've found a paper specifying that for 5G signals > 6 GHz, the channel bandwidths are 50, 100, 200 and 400 MHz. That same article lists a typical 5G cellphone receiver noise figure is expected to be around 9 dB. This leads to MDS (Minimum Detectable Signal) levels of ~ -88, -85, -82, -79 dBm.

One article I saw article suggested a required fade margin of about 7 dB.

What I have not seen is something like BER/SNR curves for the different MSC (modulation and coding) schemes. Maybe more importantly I haven't seen Packet Error Rate vs SNR curves vs MCS, or any indication of what PER / signal received signal level is required to allow a 5G cellphone to provide a usable user experience?

The genesis of my question is as follows. One person suggested one could have usable capacity with -5 dB SNR. I question whether the cellphone could provide a usable user experience at this level. Put another way, what would be a reasonable receive signal level to consider an area has 5G cellphone coverage?

Any insight, or pointers to other sites or papers will be greatly appreciated.

Thanks, Don

• You'd need to read LTE release 15 (I missed the point where that actually came out!). Fade margins really only make sense within a specific channel model – and that very specifically requires taking the wavelength into consideration. For "human ranges and speed of motion", I'd assume 7dB is too little a margin for reliability at 26 GHz – in the end, fades can be extremely close spatially at wavelengths in the 1cm regime – but I don't know whether these 7dB count for "signal power in total", "per carrier", "for the worst-case carrier"… – Marcus Müller Jun 30 '18 at 8:32
• Thanks Marcus. Looks like I'll have a lot of reading to do ... – Don Mclachlan Jul 3 '18 at 13:27