CSMA/CD was spectacularly successful for Ethernet. It enabled efficient shared-medium networking that transformed computing. When engineers began developing wireless LAN standards in the 1990s, CSMA/CD was the obvious starting point—it was proven, it was well-understood, and it was efficient.

So why didn't they just adapt CSMA/CD for wireless?

The answer lies in the fundamental physics of wireless communication. What works perfectly on a copper wire becomes physically impossible when signals travel through air. This page examines the specific technical barriers that forced IEEE 802.11 designers to abandon collision detection entirely.

Before we examine why CSMA/CD fails for wireless, let's ensure we understand exactly how it succeeds for wired networks. CSMA/CD relies on a fundamental property of shared electrical conductors: all stations can observe the same signal at effectively the same time.

Why This Works on Wire:

In wired Ethernet:

1. The medium is a conductor: Electrical signals propagate along the wire, guided by the physical structure. All stations on the segment see the same combined signal.

2. Signal levels are predictable: A normal transmission has specific voltage levels. Two simultaneous transmissions produce voltages that are the sum of both, clearly detected as abnormal.

3. The power differential is manageable: A station's own transmission on the wire and an incoming signal from another station differ by only a few dB, easily distinguishable by the receiver.

4. Propagation is nearly instantaneous: At close to the speed of light, signals traverse a 500-meter Ethernet segment in under 2.5 microseconds, allowing rapid collision detection.

The near-far problem is perhaps the most fundamental obstacle to collision detection in wireless. It describes the enormous difference in received signal strength between a station's own transmission and signals from other stations.

Understanding Signal Power Levels:

In a wireless environment, consider a station attempting to detect collisions while transmitting:

The difference between a station's own transmission and a signal from another station at typical operating range is 70-100 dB—that's a factor of 10 million to 10 billion in power!

The Path Loss Equation:

Wireless signals attenuate according to the free-space path loss equation:

Path Loss (dB) = 20 × log₁₀(d) + 20 × log₁₀(f) + 20 × log₁₀(4π/c)

Where:

• d = distance in meters
• f = frequency in Hz
• c = speed of light (3 × 10⁸ m/s)

For 2.4 GHz (802.11b/g/n):

• At 1 meter: Path loss ≈ 40 dB
• At 10 meters: Path loss ≈ 60 dB
• At 100 meters: Path loss ≈ 80 dB

Practical Implications:

Even the most sophisticated wireless receiver cannot achieve a dynamic range of 140 dB. Professional-grade receivers achieve 80-100 dB dynamic range with careful engineering. The self-signal from the transmitter would drive the receiver into saturation, making it completely blind to incoming signals.

Techniques like circulators and careful antenna isolation can reduce the self-interference by 20-30 dB, but this still leaves a 100+ dB gap—far beyond what collision detection requires.

Even if we could somehow overcome the near-far problem, collision detection would still fail due to the hidden terminal problem. This is a fundamental geometric issue inherent to wireless propagation.

The Hidden Terminal Scenario:

Consider three wireless stations: A, B, and AP (Access Point).

• Station A is on the left side of a building, 30 meters from the AP
• Station B is on the right side of the building, 30 meters from the AP
• The distance from A to B is 60 meters, with walls between them

Both A and B can communicate with the AP, but they cannot hear each other. They are "hidden" from each other's perspective.

Now imagine both A and B sense the channel idle and begin transmitting to the AP simultaneously:

1. Station A performs carrier sensing: The medium appears idle (B's signal is blocked by walls)
2. Station B performs carrier sensing: The medium appears idle (A's signal is blocked by walls)
3. Both begin transmitting
4. At the AP, both signals arrive and collide
5. Neither A nor B can detect this collision—they cannot hear each other!

Why This Breaks Collision Detection:

Collision detection requires the transmitting station to observe the collision. In the hidden terminal scenario:

• Station A transmits and monitors the channel
• Station A sees only its own signal (or noise), with no indication of B's transmission
• Station A completes its transmission, believing it succeeded
• The AP receives a corrupted frame and sends nothing (no ACK)
• Station A eventually times out and must retransmit

The collision was never detected—it was only inferred from the missing acknowledgment, which is exactly how CSMA/CA works anyway.

Collision Detection Has No Value:

In the hidden terminal case, even if Station A could somehow detect collisions during its own transmission, it couldn't detect the collision with B's signal because B's signal never reaches A. The collision detection hardware would report "no collision" while a collision is occurring at the AP.

This isn't a minor edge case—in typical deployments with walls, furniture, and varying transmission ranges, hidden terminals are common. Any MAC protocol for wireless must handle them, making collision detection not just impractical but irrelevant.

Beyond the theoretical impossibility of detecting collisions in wireless, there are practical hardware constraints that reinforce this limitation. Most wireless LAN radios are designed as half-duplex devices that physically cannot transmit and receive simultaneously.

The Transition Time:

Even switching between transmit and receive modes takes time. The T/R switch must complete its mechanical or electronic transition, the receiver front-end must stabilize, and automatic gain control must settle. This turnaround time is typically 2-5 microseconds on consumer hardware.

During this transition, the station cannot detect anything. Any collision detection scheme would have a blind spot at the critical moment when collisions are most likely—the beginning of a transmission.

Design Implications:

Because the hardware cannot detect signals while transmitting, 802.11 protocols are designed with this constraint in mind:

• All carrier sensing happens before transmission begins
• Once transmission starts, the station is committed for the entire frame duration
• Collision detection is replaced by collision inference from missing acknowledgments
• Retransmission decisions are made after complete frame transmission, not during it

Even in scenarios without hidden terminals and hypothetically full-duplex capable radios, wireless collision detection would still be unreliable due to the complex, time-varying nature of the wireless channel.

Multipath Propagation:

Wireless signals don't travel in straight lines—they reflect off walls, floors, ceilings, furniture, and even people. The receiver picks up multiple copies of the same signal, each arriving via a different path with different:

• Delay: Longer paths take more time
• Amplitude: Each reflection attenuates the signal
• Phase: Path length differences cause phase shifts

These multipath components combine at the receiver through constructive and destructive interference. The result is a received signal that can vary by 20-40 dB across distances of just half a wavelength (about 6 cm at 2.4 GHz).

Collision Detection Relies on Stable Channels:

In wired CSMA/CD, a collision creates a distinctive, detectable pattern: the signal amplitude increases beyond normal levels because two signals are now present where only one should be. The receiver can clearly distinguish "one transmission" from "two simultaneous transmissions."

In wireless, the received signal amplitude varies constantly due to fading. A 6 dB increase in signal level might be:

1. A collision (two stations transmitting)
2. Constructive interference from multipath (normal fading)
3. An obstruction moving out of the path
4. The antenna slightly rotating
5. Temperature-induced changes in the propagation environment

Without a stable reference, there's no reliable way to identify a collision by signal amplitude alone. False positives would cause unnecessary transmission aborts, while false negatives would miss actual collisions.

The Mobile Environment:

Wireless devices are often mobile. Laptops move, phones are carried, tablets are repositioned. In vehicular networks, stations move at 100+ km/h. This mobility causes rapid channel variations that completely obscure any collision signature.

An interesting phenomenon in wireless communication is the capture effect, where one signal can successfully "capture" the receiver even when another signal is present. This further complicates any collision detection scheme.

How Capture Works:

When two signals arrive at a receiver, if one is significantly stronger than the other (typically 3-10 dB difference), the stronger signal can be successfully demodulated while the weaker signal contributes only noise. The receiver "captures" the stronger signal.

This occurs because:

1. The stronger signal dominates the detector circuit
2. Error correction codes can compensate for the interference from the weaker signal
3. Modern receivers with advanced modulation (OFDM) are particularly capture-resistant but still exhibit the effect

Implications for Collision Detection:

Capture creates a complex scenario for collision detection:

From the transmitter's perspective:

• If your signal captures, you get an ACK and assume success (correct)
• If your signal is captured by another, you never know there was a collision (your signal never arrived properly)
• If both signals destroy each other, neither knows about the collision until ACK timeout

The Near-Station Advantage:

Capture creates an inherent fairness problem. Stations closer to the access point have stronger signals and are more likely to capture. Distant stations consistently lose in collisions, experiencing higher retransmission rates and lower throughput.

CSMA/CA's random backoff partially mitigates this by giving each station equal opportunity to transmit first. But the capture effect means that 802.11 networks inherently favor stations with stronger signals—a departure from the more egalitarian wired Ethernet.

Let's consolidate the fundamental differences between these two medium access protocols. Understanding this comparison is essential for appreciating why 802.11 operates the way it does.

The Efficiency Trade-off:

CSMA/CD achieves higher efficiency under light load because:

• No random backoff before first transmission attempt
• No time wasted on acknowledgments
• Collisions are detected and aborted quickly

CSMA/CA achieves better performance under heavy load because:

• Random backoff spreads transmission attempts, reducing collision rate
• Acknowledgments confirm success, enabling reliable upper-layer protocols
• Collision avoidance mechanisms scale better than collision detection

Neither protocol is universally superior—each is optimized for its operating environment.

We've examined the fundamental barriers that make CSMA/CD unsuitable for wireless networks. Let's consolidate these insurmountable challenges:

What's Next:

Now that we understand why CSMA/CA is necessary, the next page dives into the Interframe Space (IFS) mechanism in detail. We'll explore SIFS, DIFS, AIFS, and EIFS—the timing framework that enables priority-based access and orderly channel sharing in 802.11 networks.