Question bank — Wi-Fi (IEEE 802.11) — CSMA - CA, bands

False — a radio is half-duplex; its own transmitter saturates its receiver, so it hears nothing but itself. That is exactly why Wi-Fi uses avoidance (CA), not detection (CD). See CSMA-CD and Ethernet.

False — two hidden terminals both sense "idle" yet collide at the shared receiver. Physical sense only reports what you can hear, not what the receiver hears. See Hidden and Exposed Terminal Problems.

True — the NAV is a countdown set from a frame's Duration field; while NAV > 0 the station treats the channel as busy without spending power sensing (red bar in timeline 2).

False — ACKs wait the much shorter SIFS. If they waited as long as new data, a fresh sender could grab the channel and smash the ACK before it left (see the short SIFS gap and its arrow in timeline 1).

False — it only makes collisions unlikely by making two stations probably pick different counters. Two stations can still draw the same number and collide.

False — doubling increases the average wait per retry. It trades a slightly longer delay for a much lower chance of colliding again when the network is congested.

False — it freezes at its current value and resumes later. Resetting would throw away the waiting a station already did, destroying fairness.

False — a non-zero NAV counts as "busy," so the backoff counter is frozen just as if physical energy were present. Backoff only decrements when the channel is idle by both physical energy AND NAV = 0 (the red bar in timeline 2 halts everyone's countdown).

False — a success resets CW back to CWmin. Only a failure (missing ACK) doubles it. This reset is what lets the network return to short, efficient backoffs once congestion clears.

False — it must wait the longer EIFS, because the garble may have been a real frame whose ACK it couldn't read; EIFS protects that possibly-still-running exchange from being trampled.

False — it is optional and usually reserved for large frames, because the RTS+CTS overhead is wasteful on tiny ones. It is turned on mainly to fight hidden terminals.

False — 5 GHz is faster only where it reaches; its short wavelength is absorbed by walls. Band choice is a range-vs-throughput trade-off. See Bandwidth vs Throughput.

False (backwards reasoning) — it travels farther because its longer wavelength diffracts around and passes through obstacles with less attenuation. Carrying less data is a separate consequence, not the cause. See Frequency, Wavelength and Attenuation.

True — since the sender can't sense collisions, the only positive evidence of delivery is the receiver returning an ACK after SIFS. No ACK ⇒ assume loss.

False — the capture effect can let a much stronger signal be decoded correctly while the weaker one is lost. Overlap doesn't guarantee mutual destruction; it depends on relative signal strength.

The error is "listens while transmitting" — a half-duplex radio cannot. Wi-Fi never detects the collision; it infers loss from a missing ACK and then backs off.

If everyone resends at once they collide again. The fix is random backoff with an exponentially doubling window so retries spread out in time.

Incomplete — idle requires both low physical energy and NAV = 0. Virtual carrier sense can mark the channel busy even when the air is momentarily quiet (red bar in timeline 2).

Wrong — a corrupted frame is evidence someone was transmitting. The station couldn't read the Duration to set a NAV, so it defensively waits EIFS (the longest gap) instead of DIFS, giving any unheard ACK time to complete.

The real reason is spectrum: 2.4 GHz offers only ~3 non-overlapping 20-MHz channels (1, 6, 11) and is shared with microwaves, Bluetooth, etc. Crowding is about few clean channels, not popularity.

Backwards — importance means SIFS must be the shortest wait, so ACKs/CTS win the channel before any new data can start. Longer wait = lower priority, not higher.

Wrong direction — a success resets CW to CWmin; only a failure doubles it. Doubling after success would make an uncongested network needlessly slow.

This mixes up the two problems. RTS/CTS is aimed at the hidden terminal. The exposed terminal problem (a station silenced even though its transmission wouldn't actually collide) is a separate, unsolved-by-this-mechanism issue. See Hidden and Exposed Terminal Problems.

Because a transmitting radio can't simultaneously hear a faint colliding signal (half-duplex), and hidden terminals can collide at the receiver while being inaudible to each other — so detection is physically impossible.

Freezing preserves a station's "place in line" — a station that already counted down most of its wait keeps that progress, preventing starvation and giving fairness (timeline 1).

If two stations both waited through the same busy period, a fixed backoff would make them fire at the same instant and collide. Random counters almost surely differ, so one reaches zero first.

So the acknowledgement handshake completes before any station starting fresh data (which waits the longer DIFS) can seize the channel and clobber the ACK.

Because the station couldn't read the frame's Duration to build a NAV, so it doesn't know how long the real exchange lasts. EIFS is a conservative fixed wait long enough to cover a hidden ACK, avoiding trampling an exchange it can't "see."

Because the large window was a response to contention that may now be gone. Resetting lets the station return to short, efficient waits, so backoff converges back to low latency once the storm passes.

More congestion means more contending stations; a wider window offers more distinct backoff values, lowering the probability that two stations pick the same slot and collide again.

Unbounded doubling would make the delay before a retry grow without limit, effectively starving the frame. The cap bounds worst-case latency.

The hidden station C cannot hear sender A's RTS, but it can hear the access point's CTS reply. That CTS carries a Duration, so C sets its NAV and stays quiet. See Hidden and Exposed Terminal Problems.

It has huge, clean, uncrowded spectrum (more bandwidth ⇒ more data), but its short wavelength attenuates fastest through walls and air, so its reach is shortest. See Frequency, Wavelength and Attenuation.

Yes — after DIFS it still counts down a random backoff even with no contender. With CW small this is a short wait; the mechanism doesn't know it's alone.

They reach zero together and collide. Neither gets an ACK, both double their contention windows and redraw, now far more likely to differ.

The station still defers until its NAV expires — virtual sense overrides physical quiet. This is a deliberate cost that prevents mistaking a SIFS gap for the channel being free (timeline 2).

It freezes — a non-zero NAV counts as busy, so the counter neither ticks down nor resets. It only resumes once BOTH physical energy is gone AND NAV reaches 0.

It cannot set a NAV from a value it can't read, so it falls back on the physical rule and uses the conservative EIFS wait after the channel next goes idle, covering any unheard ACK before it contends.

Identically to a collision: no ACK arrives, so the sender assumes failure and enters exponential backoff. The protocol can't tell noise loss from collision loss.

Not necessarily — the capture effect lets the receiver lock onto and decode the stronger frame, ACK it, while the weaker sender gets no ACK and backs off. This quietly favours nearby/strong stations over distant ones — a fairness edge case.

$\text{Random}(0,0)=0$ (the only integer in $\{0\}$), so the station transmits immediately after DIFS with no backoff slots — the smallest possible window, offering no collision spreading at all.