Network & Telecom

Contention-Based

Q: How does contention-based access differ from contention-free access?

In contention-based access, every node competes for the same channel and decides on its own when to transmit, typically by sensing whether the medium is idle and then backing off a random interval if it is busy. There is no central authority granting time slots, so two nodes can transmit simultaneously and collide. Contention-free access, by contrast, uses a scheduler or polling master that assigns each node an explicit slot, frequency, or token, eliminating collisions entirely. Wi-Fi's Distributed Coordination Function (DCF) is contention-based; the Point Coordination Function (PCF) and 5G NR grant-based uplink are contention-free. Contention-based methods scale gracefully with bursty, unpredictable traffic and need no signaling overhead at low load, but their throughput collapses as the number of active stations rises. Scheduled methods guarantee deterministic latency and high utilization under heavy load at the cost of scheduling overhead and poor responsiveness to sporadic traffic.

Q: Why does pure ALOHA peak at only 18 percent channel efficiency?

Pure ALOHA lets a node transmit a frame the instant it has data, with no carrier sensing. A frame of duration T is destroyed if any other frame begins within the window from one frame-time before it starts to one frame-time after, a vulnerable period of 2T. Modeling arrivals as a Poisson process with offered load G frames per frame-time, the probability of zero interfering arrivals over 2T is e^(-2G), so throughput S equals G times e^(-2G). Differentiating shows the maximum occurs at G = 0.5, giving S = 0.5 times e^(-1), which is about 0.184 or 18.4 percent. Slotted ALOHA forces transmissions to align to slot boundaries, halving the vulnerable period to T; throughput becomes S = G times e^(-G), which peaks at G = 1 with S = e^(-1), about 0.368 or 36.8 percent. Carrier sensing in CSMA pushes practical efficiency far higher because nodes defer instead of blindly transmitting.

Q: What is the role of the random backoff window in CSMA/CA?

When a Wi-Fi station finds the channel busy or has just completed a transmission, it does not retry immediately; it picks a random number of idle slots from a contention window CW and counts down, transmitting only when the counter reaches zero. This randomization spreads retransmission attempts in time so that stations that deferred during the same busy period are unlikely to collide again at the same instant. After each collision the window doubles (binary exponential backoff), from CWmin of 15 slots up to CWmax of 1023 slots in 802.11. A 9 microsecond slot time means the average initial backoff is roughly 7.5 slots or about 67 microseconds. Doubling the window after collisions reduces the collision probability as the network gets busier, but it also raises latency and jitter, which is why time-sensitive traffic often migrates to scheduled or contention-free access.

/kuhn-TEN-shuhn baysd/

Describing a channel access method where every node shares one medium without a central scheduler, transmitting whenever the channel is sensed idle and resolving the resulting collisions through random access backoff rather than assigned slots. This family includes ALOHA and the CSMA/CA Distributed Coordination Function used by Wi-Fi, plus the 4-step random-access procedure that 5G NR uses to acquire an uplink grant. Contention-based schemes add almost no overhead at light load and adapt instantly to bursty traffic, but their usable throughput falls sharply as the number of competing stations grows: pure ALOHA tops out near 18% channel efficiency, slotted ALOHA near 37%, and well-tuned CSMA/CA can exceed 70% before collision rates dominate.

Category: Network & Telecom

Peak efficiency (CSMA/CA): ≈ 70 to 80%

Backoff window (802.11): 15 to 1023 slots

How Nodes Compete for a Shared Medium

A contention-based protocol replaces the idea of a master scheduler with a set of distributed rules that every node follows independently. The core decision is simple: listen before you talk. In carrier-sense multiple access (CSMA), a station monitors the channel and transmits only if it has been idle for a defined interframe space. When two stations sense an idle medium at nearly the same moment and transmit together, their signals overlap at the receiver and both frames are lost. Because radios cannot reliably hear their own transmission while sending (full-duplex on a single channel is impractical at RF power levels), wireless systems cannot detect collisions the way wired Ethernet once did. Instead they avoid collisions probabilistically, which is why the Wi-Fi mechanism is called CSMA with collision avoidance rather than collision detection.

The avoidance machinery rests on randomized timing. After the channel goes idle, each waiting station draws a random number of backoff slots from its contention window and counts down, freezing the counter whenever another station seizes the medium. Only the station whose counter first reaches zero transmits, so the random draw acts as a tie-breaker that staggers attempts in time. Every collision doubles the contention window through binary exponential backoff, which thins out the attempt rate exactly when the network is most congested. This self-tuning behavior is what lets a single Wi-Fi channel support dozens of devices, but it also injects variable latency and jitter that real-time traffic dislikes.

The cost of contention is visible in the throughput curve. As offered load rises, useful throughput climbs, peaks, and then declines as collisions and backoff consume more of the airtime than successful frames do. This is the fundamental tension that distinguishes contention-based access from contention-free scheduling, which trades flexibility for deterministic, collision-free capacity.

Throughput Models for Random Access

Pure ALOHA throughput:
S = G × e^−2G → max S ≈ 0.184 at G = 0.5

Slotted ALOHA throughput:
S = G × e^−G → max S ≈ 0.368 at G = 1

CSMA/CA average backoff (idle slots):
E[backoff] = (CW − 1) / 2 × T_slot

Binary exponential backoff window:
CW_i = min( 2ⁱ × (CW_min + 1) − 1, CW_max )

Where S = normalized throughput, G = offered load (frames per frame-time), T_slot = slot time (9 μs in 802.11n/ac), CW = contention window in slots, i = retransmission attempt. Example: CW_min = 15, T_slot = 9 μs → E[backoff] ≈ 7 × 9 = 63 μs on the first attempt.

Contention-Based vs Scheduled Access

Method	Type	Peak efficiency	Latency under load	Overhead	Best traffic profile
Pure ALOHA	Contention	≈ 18%	Unbounded	None	Sparse, tiny packets
Slotted ALOHA	Contention	≈ 37%	Unbounded	Slot sync	IoT random access
CSMA/CA (Wi-Fi DCF)	Contention	70 to 80%	Variable, high jitter	Backoff + IFS	Bursty data
TDMA / scheduled	Contention-free	90 to 98%	Bounded, deterministic	Scheduler signaling	Voice, streaming
Polling (PCF)	Contention-free	80 to 90%	Bounded	Poll frames	Managed cells

Common Questions

Frequently Asked Questions

How does contention-based access differ from contention-free access?

Contention-based nodes each decide independently when to transmit, sensing an idle channel and backing off randomly when it is busy, so simultaneous transmissions can collide. Contention-free access uses a scheduler or polling master to grant explicit slots, frequencies, or tokens, eliminating collisions. Wi-Fi DCF is contention-based; 802.11 PCF and 5G NR grant-based uplink are contention-free. Contention scales well for bursty traffic at low load but throughput collapses with many active stations, while scheduling gives deterministic latency at the cost of overhead.

Why does pure ALOHA peak at only 18 percent channel efficiency?

A pure-ALOHA frame of duration T is destroyed if any other frame starts within a vulnerable period of 2T. With Poisson offered load G, the chance of no interference is e^−2G, so throughput S = G × e^−2G, which maxes at G = 0.5 giving S ≈ 0.184. Aligning transmissions to slots (slotted ALOHA) halves the vulnerable period to T, yielding S = G × e^−G with a peak near 0.368. Carrier sensing in CSMA pushes practical efficiency far higher because nodes defer instead of transmitting blindly.

What is the role of the random backoff window in CSMA/CA?

When a station finds the channel busy or finishes a transmission, it picks a random number of idle slots from its contention window CW and counts down, transmitting at zero. Randomizing spreads retransmissions so stations that deferred together rarely collide again at the same instant. After each collision the window doubles from CWmin = 15 to CWmax = 1023 slots (binary exponential backoff). With a 9 μs slot, the average first backoff is roughly 7 slots or about 63 μs. Doubling cuts collision probability as load grows but raises latency and jitter.

Related Terms

← Container Contention-Free →