9.1 DCF, EDCA, RTS/CTS and Interframe Spaces

Every 802.11 station that wants to send a data or management frame must win the same shared medium. There is no switch fabric in the air: one OFDM or DSSS burst at a time on a channel, plus the acknowledgements that follow. Objective 3.6 is about how that sharing is scheduled — Distributed Coordination Function (DCF), Enhanced Distributed Channel Access (EDCA) / Wi-Fi Multimedia (WMM), RTS/CTS, CTS-to-Self, the Network Allocation Vector (NAV), and the interframe spaces that put acknowledgements ahead of new contenders.

This OpenExamPrep chapter teaches the IEEE 802.11 medium-access rules CWNA-109 expects you to reason about in the field. It does not claim any official relationship with CWNP; it walks the same protocol behaviors you will configure, capture, and troubleshoot.

The exam trap: collision avoidance, not collision detection

Wired half-duplex Ethernet historically used CSMA/CD (Carrier Sense Multiple Access with Collision Detection). A copper station can transmit and still notice a voltage collision on the wire, abort, and retry. A Wi-Fi radio cannot do that. While the transmitter is on, the local receiver is overwhelmed by its own energy. It cannot “listen for a collision” on the same channel in any useful way.

IEEE 802.11 therefore uses CSMA/CA — collision avoidance. Stations:

1. Sense the channel (physical and virtual).
2. Wait a mandatory idle gap (DIFS or AIFS).
3. Draw a random backoff from a contention window.
4. Transmit only when the backoff counter reaches zero.
5. Infer success or failure from the ACK (or BlockAck), not from an in-air collision signal.

If a CWNA item offers CSMA/CD as the 802.11 access method, it is describing Ethernet, not Wi-Fi. CA is the wireless rule.

Physical carrier sense versus virtual carrier sense

A station treats the medium as busy if either sense says busy. Both must be idle before a new contention even starts.

Physical carrier sense is Clear Channel Assessment (CCA) in the PHY:

• Preamble / carrier sense — the radio recognizes an 802.11 preamble and defers for the announced length of that PPDU.
• Energy detect — the radio sees RF energy above a threshold even when it cannot decode the frame (overlapping BSS, non-Wi-Fi interferer, or a PHY the chipset does not support).

Virtual carrier sense is the MAC NAV. Stations copy a microsecond count from the Duration/ID field (and related duration info in later PPDUs) into a countdown timer. While NAV is greater than zero, the station does not contend, even if CCA looks idle. RTS, CTS, data, and ACK frames all advertise remaining time so neighbors that hear only one side of an exchange still stay quiet.

Think of CCA as “do I hear RF or a preamble right now?” and NAV as “did a frame I already heard tell me the air is reserved for the next several hundred microseconds?” Hidden-node protection depends on the second question.

DCF: Distributed Coordination Function

DCF is the original distributed CSMA/CA method. QoS stations still use the same bones; EDCA only changes the idle time and the window size per queue.

Typical DCF sequence:

1. The station has an MPDU ready.
2. It senses the medium. If the medium has been idle for a DIFS, it may start backoff (or, after a long idle, some implementations transmit without an extra wait — exam items still emphasize DIFS plus backoff).
3. It draws a random integer uniformly from 0 through CW, where CW is the current contention window.
4. The backoff counter drops by one in each idle slot. If the medium goes busy, the counter freezes and resumes only after the medium is idle for DIFS again.
5. At zero, the station transmits.
6. The receiver waits SIFS and returns an ACK for unicast. Missing ACK means failure: the sender doubles CW (up to CWmax) and retries.

CW starts at CWmin (15 on many OFDM PHYs, 31 on 802.11b) and uses binary exponential backoff after failures until CWmax (often 1023). After success, CW returns to CWmin.

Worked backoff example

Two laptops both need to send after the same AP burst. Both wait DIFS. Laptop A draws backoff 4; laptop B draws 12. A transmits in slot 4. B freezes with 8 slots remaining. After A’s ACK and another DIFS, B continues from 8 rather than drawing a brand-new window. That freeze-and-resume behavior is why a station that lost once is not fully randomized out of the next round — it already paid part of the wait.

Interframe spaces: SIFS, DIFS, EIFS, and AIFS

Interframe spaces are priority by waiting. The shortest legal gap wins.

Slot time is 20 μs on long-slot DSSS/HR-DSSS and 9 μs on ERP-OFDM and later OFDM PHYs. Because DIFS and AIFS are defined from SIFS and slot time, microsecond values differ by band and PHY. CWNA cares about ranking: SIFS is always shorter than DIFS or AIFS, so an ACK is scheduled ahead of any new contender that is still waiting out its interframe space.

PIFS (SIFS + one slot) sits between SIFS and DIFS. It belongs to Point Coordination Function and some off-channel/DFS operations. Modern enterprise WLANs do not run PCF, but if a ranking question includes PIFS, the order is SIFS < PIFS < DIFS.

EIFS is the trap space. A station that failed to decode a frame does not know whether some other receiver is about to send an ACK. EIFS gives that possible ACK time to finish so the confused station does not collide with a successful exchange it only heard as noise.

EDCA and WMM access categories

Enhanced Distributed Channel Access, advertised as WMM, maps 802.1D user priority into four queues. Memorize the order high to low:

AC_VO (voice) > AC_VI (video) > AC_BE (best effort) > AC_BK (background)

Two knobs make voice statistically “win” the air more often:

1. AIFSN — voice and video typically use 2; best effort uses 3; background uses 7. After SIFS, background waits five extra slots compared with voice before it may decrement backoff.
2. Contention window — voice uses a tiny window (CWmin 3, CWmax 7 on common OFDM defaults); video is small (7/15); BE and BK keep the large 15/1023-style window.

TXOP limits also differ. VO and VI may burst several frames after winning once. BE and BK often have TXOP 0, meaning they send one frame (or one A-MPDU) and recontend.

Inside one radio, four EDCA queues contend like four virtual stations. A higher AC can “internally collide” with a lower AC on the same chip; the higher AC wins and the lower AC backs off. Across the BSS, every station’s VO queue still contends with every other VO queue. EDCA is statistical priority, not a hard TDM grant. Voice still needs capacity planning in later domains. For this objective, know which AC is highest and that AIFS and CW are the reasons.

RTS/CTS and CTS-to-Self

The hidden node problem: STA-A and STA-C both reach the AP, but not each other. Both can pass CCA, transmit together, and collide at the AP. RTS/CTS spreads NAV to both neighborhoods:

1. After winning contention, the sender transmits a short RTS. Duration covers CTS + data + ACK and the SIFS gaps.
2. Stations that hear RTS set NAV.
3. The receiver answers CTS after SIFS. Duration covers data + ACK. Stations that hear only the receiver — the hidden nodes — now set NAV too.
4. The sender transmits data; the receiver ACKs.

RTS/CTS costs airtime. Admins often set an RTS threshold so small frames skip the handshake.

CTS-to-Self is a one-frame reservation: the sender transmits a CTS with RA equal to its own address and a Duration that covers the coming payload. Neighbors that understand CTS (including 802.11b DSSS/CCK stations that cannot decode OFDM) set NAV. There is no RTS round-trip, so it is cheaper than RTS/CTS, but it only protects stations that can hear the sender. Mixed 2.4 GHz ERP cells use it as a protection tool; the next section covers when the AP turns protection on.

Putting one access attempt together

A WMM voice frame typically waits a short AIFS[VO], draws a tiny backoff, optionally runs RTS/CTS, sends data, then receives ACK after SIFS. A background backup waits a long AIFS[BK] and a large CW, so it fills leftover airtime. Both still obey NAV and CCA. A station that just heard a corrupted burst uses EIFS for that recovery interval instead of its normal DIFS or AIFS.

On the exam

• 802.11 = CA, not CD.
• Priority order is VO > VI > BE > BK.
• SIFS before ACK/CTS; DIFS for DCF; AIFS for EDCA; EIFS after a bad receive.
• NAV is virtual carrier sense; CCA is physical.
• RTS/CTS addresses hidden nodes; CTS-to-Self is a cheaper NAV reservation used heavily in ERP protection.

Key Takeaways

• Collision avoidance plus random backoff is how 802.11 shares a channel.
• Physical CCA and virtual NAV must both be idle.
• EDCA changes AIFS and CW per access category; it does not replace CSMA/CA.
• Control frames that set Duration are how stations you cannot hear still stay silent.