Noise and SNR

The idea

Every receiver hears a faint, constant hiss it can never switch off — the noise floor. It comes from the random jostling of electrons in all electronics (plus a little from the sky itself), and it is the reason radio range is finite. For a WiFi receiver the noise floor sits around −90 to −100 dBm. A signal arriving below that hiss is like a whisper at a rock concert: physically present, practically unreadable.

What matters is not how strong your signal is, but how far it stands above the hiss. That is the signal-to-noise ratio (SNR): signal level minus noise level, in dB. A signal at −60 dBm over a −90 dBm floor has 30 dB of SNR — a thousand times stronger than the noise, a clean channel. SNR is what decides your real data rate: high SNR lets WiFi use big, fast constellations like 1024-QAM; low SNR forces a retreat to slow, robust modulation like QPSK. When your speed drops at the far end of the house, it is SNR — not "bars" — doing the deciding.

Here is the cruel part: every component you add makes the hiss slightly worse. Amplifiers boost the noise along with the signal and add fresh noise of their own. How much a component degrades SNR is its noise figure, in dB — a perfect, imaginary component has a noise figure of 0 dB; a good low-noise amplifier (LNA) adds less than 1 dB; a mediocre receiver chain might add 8 dB. That is why the first amplifier after the antenna is a special low-noise design: in a receive chain, the first stage's noise figure dominates everything after it. Amplify a whisper cleanly first, and later stages can be sloppier; ruin it first, and no amount of later amplification brings the information back.

The math

SNR is a simple subtraction in dB:

$$SNR_{dB} = P_{signal}\,(\text{dBm}) - P_{noise}\,(\text{dBm})$$

The thermal noise floor itself depends on bandwidth — at room temperature:

$$P_{noise} \approx -174 + 10\log_{10}(B) + NF \quad (\text{dBm})$$

where B is the bandwidth in Hz and NF is the receiver's noise figure in dB. A 20 MHz WiFi channel gives −174 + 73 = −101 dBm before the receiver's own noise figure is added. Wider channels are faster but inherently noisier.

Common misconceptions

• Myth: Noise is interference from other transmitters. Reality: The noise floor is random thermal energy generated inside every receiver and resistor — it exists even in a perfectly quiet, shielded room. Interference is a separate, additional problem.
• Myth: More amplification fixes a weak signal. Reality: An amplifier raises signal and noise together, and adds its own noise on top. Once SNR is lost, no amplifier can restore it.
• Myth: A strong signal guarantees fast data. Reality: Data rate follows SNR, not signal strength alone. A strong signal over a strong noise floor (a crowded band, a noisy receiver) can still be slow.

Try it

Open the RF Chain Simulator and watch the noise floor rise as you add stages — then move the low-noise amplifier to the front of the chain and see the SNR recover. The Noise Figure Calculator lets you run the same cascade numbers directly.