3.1 Noise Floor, SNR, RSSI and EIRP

Why these four numbers sit on the same items

Cedar Harbor Hospital is hanging a point-to-point (PTP) radio from the main-campus roof to a parking-garage clinic 180 meters away. The facilities lead opens a laptop, sees a large received-strength reading, and says the garage radio has a great signal. Voice handsets in the clinic still clip. That sentence mixes four different ideas: how loud the wanted energy is, how loud the noise floor is, how a vendor chose to display RSSI, and how much power the transmitter is actually allowed to put into the air after antenna gain and feeder loss.

CWNA-109 Domain 1 still tests noise floor, signal-to-noise ratio (SNR), received signal strength indicator (RSSI), and equivalent isotropically radiated power (EIRP) after you already know watts, milliwatts, dB, dBm, dBi, and the rules of 10s and 3s. This OpenExamPrep section teaches those remaining measurement ideas with independent wording. CWNP writes the live exam. OpenExamPrep does not claim CWNP approval, partnership, or exact equivalence with official CWNP courseware.

Noise floor

The noise floor is the RF power present in the channel that is not the wanted 802.11 symbol you are trying to decode. It is not zero. Every receiver lives above thermal noise.

Thermal noise power in a bandwidth B is N = kTB, where k is Boltzmann's constant and T is temperature in kelvin. For a 20 MHz WLAN channel at room temperature, kTB is about -101 dBm. A real radio then adds a noise figure: extra noise from low-noise amplifiers, mixers, and analog-to-digital conversion. A 6 dB noise figure raises that quiet-lab floor to about -95 dBm. The environment then adds more: neighboring BSS energy, Bluetooth, poorly shielded motors, lighting ballasts, and non-802.11 emitters. In a busy hospital 2.4 GHz channel, a spectrum analyzer may show a noise-plus-interference floor closer to -85 dBm or worse even when your own AP is quiet.

Bandwidth is part of the physics. Double the channel width and you collect about 3 dB more thermal noise in the kTB sense: 40 MHz is roughly 3 dB louder than 20 MHz, 80 MHz another 3 dB, if the radio's noise figure does not change. That is one reason a wide channel can look hot on a spectrum analyzer even when no single interferer is huge.

Exam-useful facts:

• The noise floor is a power, usually in dBm, not a percentage and not a bar graph.
• You cannot invent SNR by cranking a receive reporting slider. If noise rises with the wanted energy, the ratio stays poor.
• A high noise floor can be on-channel 802.11 (co-channel contention), adjacent-channel splatter, or non-802.11 energy. This chapter only needs you to treat that energy as the noise term. Later validation chapters name the sources.

At Cedar Harbor, the garage radio sits beside a rooftop HVAC variable-frequency drive. The drive sprays broadband energy into 2.4 GHz. The noise floor on that radio is high even at 3 a.m., when client count is zero.

Signal-to-noise ratio

SNR is how far the wanted signal sits above the noise (and, in the field, above noise-plus-interference). In decibels:

SNR (dB) = signal power (dBm) − noise power (dBm)

If the wanted energy is −62 dBm and the noise floor is −92 dBm, SNR is 30 dB. If both are 10 dB stronger, SNR is still 30 dB. That is why a stronger RSSI is not automatically a better link.

SNR, not raw amplitude, is what lets a demodulator pick the correct constellation point. Higher-order modulation needs more SNR than BPSK or QPSK. Voice and high-MCS data therefore fail first when SNR collapses, even if a status LED still looks green. A practical indoor WLAN often aims for roughly 25 dB SNR or better for data-heavy SSIDs; real-time voice designs usually ask for still more margin. Exact design targets belong in RF-validation chapters. For this section, lock the definition and the subtraction.

SNR is a ratio. It is reported in dB, not dBm. Mixing the units on an exam item is a classic miss: dBm is an absolute power; dB is a difference.

RSSI is vendor-relative, not a standardized quality metric

RSSI is a received signal strength indicator. IEEE 802.11 left the scale implementation-specific. Vendors may report:

• a value that looks like dBm (for example −48),
• a 0–100 or 0–60 integer,
• or a chipset RSSI that is not calibrated to a laboratory power meter.

Two radios from two vendors, on the same cart, hearing the same frame, can show different RSSI numbers. You cannot treat RSSI as a standardized quality metric, and you cannot compare RSSI across brands as if they shared a calibration. RSSI also does not tell you noise. A radio can report a strong RSSI while the channel is loud with interference.

When a vendor does present receive power in dBm on that same radio, you may use it as the signal term in SNR. That still does not make RSSI a universal score you can paste into a survey report next to another vendor's integer.

Exam trap: strong RSSI with poor SNR. Cedar Harbor's garage client shows RSSI −45 (excellent-looking) because it sits 8 meters from the garage AP. The HVAC noise floor is −55 dBm. SNR is only about 10 dB. Frames retry, MCS falls, and voice clips. The laptop great signal was RSSI, not SNR.

The subtraction that ties the three together

When RSSI and the noise floor are both in dBm on the same radio:

SNR ≈ RSSI − noise floor

Example: RSSI = −62 dBm, noise = −90 dBm → SNR ≈ 28 dB.

If RSSI is a 0–100 vendor score, do not subtract. Convert nothing. Use that vendor's own SNR field, a spectrum analyzer, or a radio that actually reports dBm.

EIRP: what actually leaves the antenna

Conducted transmit power is what the radio delivers at its connector. The antenna then adds passive gain (focusing energy, taught in the next section). Cables, pigtails, lightning arrestors, and connectors subtract loss. Equivalent isotropically radiated power (EIRP) is the power you would need from a lossless isotropic antenna to produce the same power density in this antenna's main beam.

EIRP (dBm) = Tx power (dBm) + antenna gain (dBi) − cable and connector loss (dB)

Use dBi for the gain term. Loss is in dB. Do not mix dBd (dipole-referenced gain) unless you convert: dBi ≈ dBd + 2.15. Some older catalogs print dBd; adding that number as if it were dBi understates EIRP by about 2 dB.

Worked example (memorize this arithmetic):

• Radio conducted power: 20 dBm
• Antenna gain: 6 dBi
• Cable and connector loss: 2 dB

EIRP = 20 + 6 − 2 = 24 dBm

24 dBm is about 251 mW (20 dBm = 100 mW; +3 dB ≈ ×2 → 23 dBm ≈ 200 mW; +1 dB more is about 251 mW). If Cedar Harbor uses a longer thin jumper that actually loses 4 dB instead of 2 dB, EIRP falls to 22 dBm. If someone tries to fix coverage by swapping a 6 dBi omni for a 10 dBi omni without lowering conducted power, EIRP rises 4 dB and may violate a regulatory cap.

A lightning arrestor is another series loss. If the arrestor is 0.5 dB and two N-connectors are 0.25 dB each, that 1 dB belongs in the same subtraction. People who only add radio + antenna and ignore the feeder invent an EIRP that the far end will never see.

Regulatory EIRP caps are not one worldwide number

Regulators typically limit EIRP—or conducted power plus a stated antenna-gain allowance—because EIRP is what neighboring systems actually experience. Limits vary by:

• regulatory domain (United States FCC tables versus ETSI Europe versus other national tables),
• band (2.4 GHz, 5 GHz UNII sub-bands, 6 GHz indoor versus standard-power classes),
• use (indoor AP, outdoor AP, point-to-multipoint, fixed point-to-point),
• device class (AP versus client; low-power indoor versus standard power).

Do not memorize a single invented phrase such as the FCC EIRP is always this one number for every Wi-Fi radio. As one United States 2.4 GHz point-to-multipoint illustration, many treatments of FCC Part 15.247 start from 30 dBm (1 W) conducted with 6 dBi of antenna gain before you must reduce conducted power for extra gain, which people quote as 36 dBm EIRP for that PTMP case. ETSI Europe commonly caps 2.4 GHz at 20 dBm EIRP—a much lower ceiling. 5 GHz UNII sub-bands and 6 GHz classes use different tables, and point-to-point rules can allow higher-gain antennas with power-backoff formulas that are not the PTMP formula. Certified equipment must use the domain and band programmed on the radio. Cedar Harbor's U.S. campus and a sister clinic operating under ETSI rules cannot share one EIRP cheat-sheet.

If EIRP is too high, you are not extra covered—you are out of spec and you raise interference for everyone else. If EIRP is too low because of a long, thin, high-frequency jumper, the far end shows weak RSSI and you still need enough SNR.

Putting it together on the garage link

Cedar Harbor's roof radio: 20 dBm conducted, 6 dBi patch, 2 dB of N-connectors and jumper → 24 dBm EIRP. The garage radio reports RSSI −52 dBm and noise −72 dBm on the same dBm scale → SNR ≈ 20 dB. That SNR, not the bars, decides whether MCS holds. A firmware screen that shows only RSSI would have hidden the HVAC problem.

Change the feeder to a 12-meter thin jumper with 5 dB of loss at 5 GHz. Same radio, same 6 dBi antenna: EIRP = 20 + 6 − 5 = 21 dBm. The far-end RSSI drops about 3 dB if nothing else changes. If the noise floor stays put, SNR drops about 3 dB too. Cable loss is not a cosmetics issue; it is in the EIRP line.

Measurement traps

1. Treating RSSI as a universal quality score.
2. Comparing RSSI from two vendors as if they shared a calibration.
3. Ignoring noise when RSSI looks strong.
4. Forgetting cable, arrestor, and connector loss in the EIRP sum.
5. Adding dBi gain and forgetting to subtract loss.
6. Claiming one EIRP number for every country and band.
7. Mixing dBd and dBi in the EIRP line.
8. Subtracting a 0–100 RSSI from a dBm noise floor.