At 290 K, estimate receiver sensitivity from -174 dBm/Hz + 10 log10(B in hertz) + cascaded noise figure + the required SNR, with any implementation or measurement margin stated separately. For coded or spread waveforms, use the correct Eb/N0 or Es/N0 relationship and processing gain rather than substituting a generic SNR. Define the receiver input plane, bandwidth, waveform, error-rate target, gain state and temperature.
For a 20 MHz receiver, thermal noise is approximately -101.0 dBm. Adding 5 dB noise figure, 12 dB required SNR and 2 dB implementation margin gives about -82.0 dBm. Doubling the noise bandwidth to 40 MHz raises integrated thermal noise by 3 dB, so sensitivity worsens by about 3 dB if noise figure and required SNR stay unchanged.
Do not use this arithmetic alone as an installed sensitivity guarantee. Loss ahead of the first LNA, mismatch, filter ripple, temperature, blocker-induced compression, LO phase noise, ADC range and gain-control behavior can move the result. Reject a sensitivity claim that omits bandwidth or error criterion, or that was measured without the blockers and duplex leakage present in operation. Supply the wanted waveform, receiver bandwidth, input losses, NF by gain state, required BER/BLER/PER or EVM, blockers, maximum input, recovery and test uncertainty.