Intuition What this page isThe parent note told you the rules of CSMA/CA and the bands. This page makes you run those rules by hand on every kind of situation an exam (or real router) can throw at you. We list every case class first, then work one example per cell until the whole grid is covered.
Every question about this topic falls into one of these cells. We will hit all of them.
#
Cell (case class)
What makes it tricky
Covered by
A
Two stations contend, both count down who wins, who freezes
Ex 1
B
Backoff tie (both draw the same number)the degenerate case → real collision
Ex 2
C
Exponential backoff growth to the capdoubling formula, min clamp
Ex 3
D
Timing a full exchange (DIFS+backoff+data+SIFS+ACK)adding all the gaps in µs
Ex 4
E
Hidden terminal WITHOUT RTS/CTS why carrier sense fails
Ex 5
F
Hidden terminal WITH RTS/CTS + NAV how the fix silences the hidden node
Ex 6
G
Band / range word problem (real world)frequency ↔ wavelength ↔ penetration
Ex 7
H
Channel-crowding on 2.4 GHz (exam twist)why only 3 clean channels
Ex 8
I
Degenerate limits (C W = 0 zero-slot backoff, overhead
Ex 9
J
Exposed terminal (the mirror-image of hidden)when a station stays quiet needlessly
Ex 10
Terms used below, all built in the parent note, restated in one line so you never guess:
slot time = the small fixed clock tick (e.g. 9 μ s 20 μ s C W C W m i n initial (smallest) contention window a fresh frame uses — the 802.11 default is 15 C W m a x cap the window may not exceed no matter how many collisions occur — a common value is 1023 backoff counter = a random whole number of slots you must sit idle before you may talk.SIFS / DIFS = fixed waiting gaps; SIFS is short (for replies), DIFS is long (for new data).NAV = a countdown timer that says "channel is reserved, stay quiet" — a virtual busy signal.
Definition Where does the defaultC W = 15 whole slots numbered 0 C W . The 802.11 standard sets the initial window C W m i n { 0 , 1 , … , 15 } 16 possible slots . Hardware likes powers of two, and 16 = 2 4 maximum draw is 16 − 1 = 15 16 slots, top index 15 " is exactly why the number 15 C W m a x 1023
Prereqs if any line feels shaky: CSMA-CD and Ethernet , MAC Layer and ARQ / ACK , Hidden and Exposed Terminal Problems , Frequency, Wavelength and Attenuation , Bandwidth vs Throughput .
Worked example Example 1 — X vs Y, different drawsChannel is busy. Station X and station Y each have a frame ready. When the channel goes idle both wait DIFS , then draw a backoff:
X = 4 Y = 7 = 9 μ s
Forecast: guess — will Y ever get a turn, or starve forever?
Both wait DIFS after the channel goes idle.
Why this step? New data must wait the long gap (DIFS) so it can never jump ahead of a reply (which uses the shorter SIFS).Both decrement once per idle slot. After 4 idle slots, X's counter reaches 0.
Why this step? The counter counts idle slots only; the smaller number reaches 0 first.X transmits. Y has counted 4 7 − 4 = 3 Why this step? Y saw the same 4 idle slots, so it dropped by 4.Y senses the channel busy (X is sending) and FREEZES at 3.
Why this step? Freeze, not reset — this keeps Y's place in line (fairness).Next round: channel idle again → DIFS → Y resumes from 3 , a head start, so Y almost surely wins next.
Why this step? A frozen low counter beats a freshly-drawn random counter most of the time → no starvation.
Verify: X's countdown length = 4 × 9 = 36 μ s = 7 − 4 = 3 = 27 μ s 3 > 0
Intuition How to read Figure s01The orange line is X's counter, the teal line is Y's. Both slope down together (same idle slots seen). Watch orange reach the black zero-line at slot 4 → X transmits; at that instant the plum band opens (channel busy) and teal stops moving, frozen at 3. The picture is the "freeze, not reset" rule.
Worked example Example 2 — Same draw → real collisionSame setup, but this round both draw 5 . slot time = 9 μ s { 0 , … , 15 }
Forecast: does CSMA/CA guarantee no collision, or only make it unlikely ?
Both wait DIFS, both count down from 5.
Why this step? Identical draws mean identical countdowns.After 5 idle slots BOTH hit 0 in the same slot → both transmit → collision.
Why this step? Nothing separates them; carrier sense sees "idle" for both until the instant they both fire.Neither receives an ACK (frames overlapped and were destroyed).
Why this step? No ACK is the only collision evidence Wi-Fi has (a radio can't hear the smash directly).Both double C W Initial C W = 15 C W = 2 ( 15 + 1 ) − 1 = 31 Why this step? A wider window makes a repeat tie less likely.
Verify: collision slot for both = 5 × 9 = 45 μ s 2 ( 15 + 1 ) − 1 = 31 not eliminate, collision probability. ✓
Worked example Example 3 — Doubling until it clampsStart C W = 15 C W m a x = 1023 every attempt. List C W
Forecast: how many collisions until C W
Apply C W new = min ( 2 ( C W old + 1 ) − 1 , 1023 ) Why this step? Each failed attempt widens the window to spread retries.Compute:
15 → 31 → 63 → 127 → 255 → 511 → 1023 → 1023 Why this step? Each term is 2 ( prev + 1 ) − 1 min clamps it.
After 6 doublings C W Why this step? The cap prevents infinite delay — unbounded backoff = a station that never talks.
Verify: sequence check 2 ( 15 + 1 ) − 1 = 31 2 ( 31 + 1 ) − 1 = 63 2 ( 511 + 1 ) − 1 = 1023 min ( 2 ⋅ 1024 − 1 , 1023 ) = 1023
Intuition Where do the "+1" and "−1" come from? (the magic constants)The window value C W one less than the number of slots (slots = C W + 1 0 double the number of slots , not the index. So:
C W old + 1 slot count (e.g. 15 → 16 2 × 16 → 32 − 1 index (32 → 31
In short: index → +1 → slots → ×2 → slots → −1 → index. The + 1/ − 1 × 2
Intuition How to read Figure s02Each plum dot is C W orange dashed line at 1023 min clamp doing its job. The curve visibly shows "grow fast, then stop."
Worked example Example 4 — Add up one successful transferGiven: SIFS = 10 μ s DIFS = 28 μ s = 9 μ s = 200 μ s = 24 μ s = 3
Forecast: which single gap contributes the most time here?
DIFS wait = 28 μ s Why this step? Data must wait the long IFS before touching backoff.Backoff = 3 × 9 = 27 μ s Why this step? Count down 3 idle slots to 0, then transmit.Data frame = 200 μ s Why this step? Actual payload air time.SIFS = 10 μ s Why this step? The receiver replies after the short gap so nothing pre-empts the ACK.ACK = 24 μ s Why this step? The proof of delivery arrives.Total = 28 + 27 + 200 + 10 + 24 = 289 μ s Why this step? Every phase is sequential — just add.
Verify: 28 + 27 + 200 + 10 + 24 = 289 μ s 200 μ s shrinking IFS/backoff barely helps efficiency once frames are large — a real design insight. ✓
Intuition How to read Figure s03Time flows left to right; each coloured block is one phase and its width is drawn to scale in µs. Notice the teal DATA block dwarfs everything else — that visual dominance is exactly the "overhead barely matters for big frames" insight. Add the block widths to get the 289 μ s
Worked example Example 5 — Carrier sense failsStations A and C can both reach access point B , but A and C are too far apart to hear each other. Both have data. No RTS/CTS. What happens?
Forecast: will plain carrier sense save them?
A senses the channel: idle (it cannot hear C at all). A transmits.
Why this step? Physical carrier sense only reports what A can hear.C senses the channel: also idle (it cannot hear A). C transmits.
Why this step? Same blind spot from C's side.At B, the two frames overlap → collision at the receiver.
Why this step? Collisions happen at the listener , not between two deaf talkers.B sends no ACK; both A and C time out and back off.
Why this step? Missing ACK is the only signal — but the wasted air time already cost throughput.
Verify: both saw "idle" yet collided → carrier sense alone is insufficient when terminals are hidden. This motivates the NAV fix in Ex 6. ✓ (See Hidden and Exposed Terminal Problems .)
Intuition How to read Figure s04A (orange) and C (plum) sit at the two ends; B (teal) is the access point in the middle. The two arrows both point inward to B and meet there — the red "COLLISION at B" label marks the exact spot where the frames overlap. The point is that the damage happens at the receiver , not between the two senders.
Worked example Example 6 — The fix silences the hidden nodeSame A / B / C, but now A uses RTS/CTS . RTS says "I need the channel for 300 μ s
Forecast: C never heard A — how does C ever learn to shut up?
A sends a tiny RTS to B carrying a Duration = 300 μ s Why this step? A cheap probe to reserve the channel before the big frame.B replies with CTS , also carrying the Duration, after SIFS.
Why this step? CTS is broadcast around B — and C can hear B even though it can't hear A.C hears the CTS → sets its NAV = 300 μ s for that whole time.
Why this step? Virtual carrier sense: C trusts the announced Duration, no physical sensing needed.A sends its data collision-free; B ACKs.
Why this step? The channel is reserved, so the hidden collision cannot occur.
Verify: C's silence window = 300 μ s avoid the large collision — worth it only for big frames. ✓
Intuition How to read Figure s05Follow the arrows in order: orange RTS from A reaches only B; then the teal CTS spreads from B outward and reaches C . That single teal arrow is the whole fix — it is the bridge that carries the reservation across the gap A could never cross. The plum note under C ("NAV=300us, STAYS SILENT") is the result.
Worked example Example 7 — Which band reaches the far room?Your router does 2.4 GHz and 5 GHz. Speed of a radio wave c = 3 × 1 0 8 m/s
Forecast: which wavelength is longer — and does longer help or hurt penetration?
Wavelength λ = c / f Why this step? Wavelength is the physical "size" of the wave; it governs how it bends around and passes through obstacles. (See Frequency, Wavelength and Attenuation .)2.4 GHz: λ = 2.4 × 1 0 9 3 × 1 0 8 = 0.125 m = 12.5 cm Why this step? Plug f = 2.4 × 1 0 9 Hz 5 GHz: λ = 5 × 1 0 9 3 × 1 0 8 = 0.06 m = 6 cm Why this step? Plug f = 5 × 1 0 9 Hz Longer wave (2.4 GHz) diffracts around and loses less energy through walls → reaches the far room.
Why this step? Bigger waves are absorbed/blocked less; higher frequency carries more data but dies faster.
Verify: 3 × 1 0 8 /2.4 × 1 0 9 = 0.125 m 3 × 1 0 8 /5 × 1 0 9 = 0.06 m 0.125 > 0.06
Intuition How to read Figure s06The orange wave (2.4 GHz) is stretched out (long wavelength); the teal wave (5 GHz) is squished (short wavelength). The brown vertical "wall" band lets the long orange wave keep going on the far side, while the teal wave shrinks to almost nothing — a picture of "long waves penetrate, short waves get absorbed."
Worked example Example 8 — Why only 3 clean channels?The 2.4 GHz ISM band spans about 2.400 2.4835 GHz ≈ 83.5 MHz 20 MHz 5 MHz non-overlapping channels fit?
Forecast: guess before computing — is it 4? 11? 3?
Two channels don't overlap only if their centres are ≥ 20 MHz
Why this step? Each occupies 20 MHz Bandwidth vs Throughput ).In theory 20 MHz = 20/5 = 4 just avoid overlap — so channels 1 , 5 , 9 , 13 Why this step? This is the naive minimum-spacing answer many students give.In practice real radios leak a little energy past their nominal 20 MHz 5-number (25 MHz) spacing → the classic 1 , 6 , 11 Why this step? The extra channel-number of gap buys a safety margin against real-world spectral leakage, at the cost of one fewer usable channel.Count 25 MHz 83.5 MHz ⌊ 83.5/25 ⌋ = 3 1 , 6 , 11 3 truly clean ones.
Why this step? Only ~3 non-overlapping slots fit with the safe spacing → the band is crowded .Add microwaves, Bluetooth, baby monitors sharing the band → real throughput drops further.
Why this step? Explains the parent note's "2.4 GHz is crowded" claim.
Verify: naive 20/5 = 4 5 1 → 6 → 11 ⌊ 83.5/25 ⌋ = 3
Worked example Example 9 — The edge cases(a) A station draws backoff = 0 54 Mbps 24 Mbps
Forecast: if backoff is 0, does the station skip all waiting?
Backoff = 0 the station still waited DIFS in step 1; a 0 counter just means transmit immediately after DIFS , no idle slots.
Why this step? Random ( 0 , C W ) includes 0; DIFS is separate and always paid.So even the minimum-backoff station is not collision-proof — another station could also draw 0 (the Ex 2 tie).
Why this step? 0 is just the smallest of many equally-likely values.Throughput = 24 Mbps = 54 Mbps efficiency = 24/54 ≈ 0.444 = 44.4% Why this step? Bandwidth is the raw rate; throughput is delivered data after all the CSMA/CA overhead (see Bandwidth vs Throughput ).The missing ∼ 56% is DIFS + backoff + SIFS + ACK + headers — pure coordination cost.
Why this step? Shows overhead is the price of sharing air politely.
Verify: 24/54 = 0.4 4 ≈ 44.4% not a free instant send. ✓
Worked example Example 10 — Staying quiet for no reasonFour stations in a row: B wants to send to A ; C wants to send to D . C can hear B's transmission, but A is out of C's range and D is out of B's range. Should C wait?
Forecast: C hears the channel busy — is holding back the right move?
B transmits to A. C's physical carrier sense reports "busy" (C hears B).
Why this step? Carrier sense only knows someone nearby is talking , not who they are talking to .Naive carrier sense makes C defer , even though C's target D is nowhere near A.
Why this step? This is the exposed terminal problem — C is exposed to B's signal and needlessly silenced.The two transmissions B→A and C→D would NOT have collided — A only hears B, D only hears C.
Why this step? Collisions are decided at the receivers (A and D), and neither receiver hears the other sender.Cost: wasted opportunity — C could have sent in parallel. RTS/CTS helps the hidden case but can worsen the exposed case, so real designs treat this as a known trade-off.
Why this step? Shows hidden and exposed are opposite failure modes: hidden = collide when you shouldn't be allowed , exposed = stay silent when you should be allowed .
Verify: receivers A and D are mutually out of range → parallel B→A and C→D succeed; C's deferral was unnecessary. Exposed = lost throughput, not a collision. ✓ (See Hidden and Exposed Terminal Problems .)
Intuition How to read Figure s07Read the row B–A–…–C–D. The green arrows B→A and C→D point away from each other; their receivers (A and D) sit at the far outer ends, each in range of only its own sender (dashed range circles). Because the arrows never share a listener, both could run at once — yet C hears B and freezes, the wasted-opportunity that defines the exposed terminal.
Recall Cover the answers and test each cellCell A — after X (draw 4) transmits, where is Y (draw 7)? ::: Frozen at 3.
Cell B — do equal backoff draws collide? ::: Yes — CSMA/CA only lowers collision odds, never zero.
Cell C — C W
Mnemonic The exchange orderD IFS → B ackoff → D ata → S IFS → A CK = "D on't B e D umb, S end A cks."
Mnemonic Hidden vs ExposedHidden = you can't hear the other sender, so you collide (talk when you shouldn't). Exposed = you can hear a sender aimed elsewhere, so you stay silent (quiet when you needn't). Opposite mistakes, same root: carrier sense judges the sender , but collisions live at the receiver .
