Loading learning content...
Consider three approaches to managing access to a shared conference room:
Approach 1: "First Come, First Served with Retry" Anyone can walk in when it looks empty. If someone else is already there, you leave and try again later. Simple and fast when the room is usually free, but chaotic when many people want it.
Approach 2: "Scheduled Coordination" A token (like a talking stick) circulates. Only the person holding the token can use the room. No conflicts, completely fair, but you might wait a while even if nobody else actually needs the room.
Approach 3: "Time Slots" Each person is assigned specific hours. You get guaranteed exclusive access during your slot, but if you don't need the room during your time, it sits empty.
These three approaches map directly to the three major categories of MAC protocols:
This page provides comprehensive coverage of each category, preparing you to understand specific protocols in subsequent modules.
By the end of this page, you will understand the design philosophy, advantages, disadvantages, and appropriate use cases for each MAC protocol category. You'll be able to classify any MAC protocol into its category and predict its behavior characteristics based on that classification.
Random access protocols (also called contention-based protocols) allow any station to transmit whenever it has data, without coordination or permission from other stations. The term "random" refers to the unpredictable nature of transmission opportunities—they arise based on local conditions and randomized delays, not on a predetermined schedule.
The fundamental philosophy of random access is optimistic transmission:
"Channel appears idle? Go ahead and transmit. If it doesn't work out (collision), we'll recover."
This philosophy assumes that:
No Central Coordination: Stations make independent decisions. There's no controller, scheduler, or master station determining who transmits when. This makes random access protocols distributed and decentralized.
Collision-Tolerant Design: Since collisions are expected, the protocol includes mechanisms for:
Load-Dependent Performance: Random access protocols perform very differently under different loads:
| Protocol | Key Innovation | Maximum Efficiency | Use Case |
|---|---|---|---|
| Pure ALOHA | Simple random access | 18.4% | Historical; conceptual foundation |
| Slotted ALOHA | Time synchronization | 36.8% | Satellite, cellular uplink |
| CSMA (non-persistent) | Carrier sensing + backoff | ~80% | Conceptual; leads to CSMA/CD |
| CSMA (1-persistent) | Carrier sensing + immediate transmit | ~50% | Conceptual |
| CSMA (p-persistent) | Carrier sensing + probabilistic | Variable | Conceptual; tunable |
| CSMA/CD | Collision detection + early abort | 95%+ | Traditional Ethernet |
| CSMA/CA | Collision avoidance | ~70% | Wi-Fi, wireless LANs |
Generation 1: Pure ALOHA The simplest approach: transmit whenever you have data. If no acknowledgment arrives, wait a random time and retransmit. Terrible efficiency (18.4% max) but proved random access was viable.
Generation 2: Slotted ALOHA Synchronize all stations to time slots; transmit only at slot boundaries. Doubles theoretical efficiency to 36.8% by eliminating partial collisions.
Generation 3: CSMA (Carrier Sense Multiple Access) Listen before transmitting. If the channel is busy, don't transmit (wait for idle). This dramatically reduces collisions because stations avoid obvious conflicts. Variants (1-persistent, non-persistent, p-persistent) differ in how aggressively they attempt transmission when the channel becomes idle.
Generation 4: CSMA/CD (Collision Detection) Add collision detection: if a collision is happening, stop transmitting immediately rather than waste time completing a doomed frame. Abort, send jam, backoff, retry. This enables Ethernet's 95%+ efficiency.
Generation 5: CSMA/CA (Collision Avoidance) For wireless where detection is impossible, add avoidance mechanisms: RTS/CTS reservation, mandatory inter-frame gaps, virtual carrier sensing via NAV. Accept lower efficiency for wireless compatibility.
Random access excels when traffic is bursty (long idle periods between transmissions), station count is variable, low latency under normal conditions matters more than guaranteed latency, and simplicity is valued. This describes most LAN and Wi-Fi scenarios.
Controlled access protocols take the opposite approach from random access: they coordinate station transmissions so that collisions are impossible by design. Rather than letting stations compete and recovering from failures, controlled access grants explicit permission before any station transmits.
The fundamental philosophy of controlled access is coordinated exclusivity:
"Only one station has permission to transmit at any moment. Permission is passed in an orderly fashion. Collisions cannot occur."
This philosophy assumes that:
Explicit Permission Required: A station can only transmit when it holds a token, receives a poll from a controller, or has a reserved slot. No unsolicited transmissions occur.
Zero Collisions: Since only one station can ever transmit at a time, frames never overlap. Every transmission succeeds (assuming no physical layer errors).
Deterministic Access: The maximum time a station must wait for its transmission opportunity is bounded. This is critical for real-time applications that cannot tolerate unbounded delay.
Overhead Under Light Load: Unlike random access (which is very efficient when traffic is light), controlled access has fixed overhead for the coordination mechanism—even when no one wants to transmit.
How it works:
Characteristics:
Optimization: Select-reject polling (poll only stations with pending data), group polling, priority polling.
Use cases: Industrial control systems (Modbus), mainframe terminal networks (historic), sensor networks with strict timing requirements.
How it works:
IEEE 802.5 Token Ring:
Token Bus (IEEE 802.4):
| Characteristic | Polling | Token Ring | Token Bus |
|---|---|---|---|
| Coordination | Centralized controller | Distributed token | Distributed token |
| Physical topology | Star or multidrop | Ring | Bus |
| Single point of failure | Controller | Monitor station | Token holder (recoverable) |
| Empty station overhead | Poll-response time | Token pass only | Token pass only |
| Priority support | Poll order | Token holding priority | Token rotation priority |
| Used in | Industrial, legacy terminals | LANs (historical) | Factory automation (historical) |
Controlled access excels when bounded latency is required (real-time control, audio/video), traffic is consistently heavy, fairness must be guaranteed, or the application cannot tolerate any collisions. Industrial automation and traditional telecommunications often use controlled access.
Channelization protocols (also called multiple access protocols in cellular terminology) solve the sharing problem by dividing the channel itself. Rather than coordinating access to a single shared channel, the channel's capacity is partitioned so that each station has its own dedicated sub-channel.
The fundamental philosophy of channelization is resource partitioning:
"Divide the total channel capacity into pieces. Give each station its own piece. Stations use their pieces independently without interference."
This philosophy assumes that:
Pre-Allocated Resources: Each station receives a guaranteed portion of channel capacity. This allocation may be fixed or dynamically adjusted.
No Contention: Since stations use different sub-channels, they never interfere with each other. There are no collisions and no need for collision handling.
Potential Waste: If a station has no data to send, its allocated sub-channel sits idle. The capacity cannot be used by other stations (in pure channelization).
Rigid Partitioning: The division is based on fixed physical properties: frequency bands (FDMA), time slots (TDMA), or orthogonal codes (CDMA).
How it works: The available frequency spectrum is divided into non-overlapping bands. Each station is assigned one (or more) bands for exclusive use.
| Band 1 | Guard | Band 2 | Guard | Band 3 | Guard | Band 4 |
| Station A| Band | Station B| Band | Station C| Band | Station D|
Characteristics:
How it works: Time is divided into frames, each containing slots for each station. Stations transmit only during their assigned slots.
Frame 1: |Slot A|Slot B|Slot C|Slot D|Guard|
Frame 2: |Slot A|Slot B|Slot C|Slot D|Guard|
Frame 3: |Slot A|Slot B|Slot C|Slot D|Guard|
Characteristics:
How it works: Each station multiplies its data by a unique spreading code (a long pseudo-random sequence). All stations transmit simultaneously on the same frequency. Receivers multiply the received signal by the correct code to extract the desired data.
The magic of orthogonal codes: When spreading codes are orthogonal (mathematically independent), the receiver's multiplication process filters out signals from other stations—they appear as low-level noise.
Characteristics:
| Dimension | FDMA | TDMA | CDMA |
|---|---|---|---|
| Separation method | Frequency bands | Time slots | Orthogonal codes |
| Simultaneous users | Separated in frequency | Separated in time | All share frequency and time |
| Guard requirement | Guard bands (frequency) | Guard times (time) | Power control (amplitude) |
| Synchronization | Not required | Critical (tight) | Less critical (codes) |
| Wasted capacity | Idle bands | Idle slots | Reduced capacity (not wasted) |
| Scalability | Limited by spectrum | Limited by frame size | Soft limit (graceful) |
| Complexity | Low | Medium | High (spread spectrum) |
Modern systems dynamically allocate channels:
This dynamic multiple access captures the guaranteed-access benefits of channelization while avoiding the waste of fixed allocation.
Channelization excels when stations have continuous traffic (voice, video streaming), guaranteed Quality of Service is required, the number of stations is known and stable, or the application cannot tolerate any contention delay. Cellular networks and satellite links commonly use channelization.
Understanding when to use each MAC protocol category is essential for network design. The choice depends on traffic characteristics, performance requirements, and practical constraints.
Bursty traffic (web browsing, file transfers): Random access excels. Stations transmit occasionally but need low delay when they do. Channelization would waste most allocated capacity during idle periods.
Continuous traffic (voice, video): Channelization shines. Dedicated sub-channels provide guaranteed bandwidth without contention. Controlled access also works well.
Mixed traffic: Hybrid approaches or dynamic allocation. Many modern systems combine categories—for example, LTE uses scheduled access for data but contention-based access for initial connection.
Minimum delay under light load: Random access wins. Transmit immediately when channel is free. Controlled access and channelization introduce overhead even when unnecessary.
Bounded worst-case delay: Controlled access or channelization required. Random access has unbounded worst-case delay (theoretically infinite retries).
Maximum throughput under heavy load: Controlled access or channelization. Random access collapses under overload; the other categories maintain performance.
| Requirement | Random Access | Controlled Access | Channelization |
|---|---|---|---|
| Low delay (light load) | ★★★★★ | ★★☆☆☆ | ★★★☆☆ |
| Low delay (heavy load) | ★☆☆☆☆ | ★★★★☆ | ★★★★★ |
| Bounded latency | ☆☆☆☆☆ | ★★★★★ | ★★★★★ |
| Bursty traffic efficiency | ★★★★★ | ★★★☆☆ | ★★☆☆☆ |
| Continuous traffic efficiency | ★★☆☆☆ | ★★★★☆ | ★★★★★ |
| Implementation simplicity | ★★★★★ | ★★★☆☆ | ★★★☆☆ |
| Dynamic station count | ★★★★★ | ★★★☆☆ | ★★☆☆☆ |
| Guaranteed bandwidth | ☆☆☆☆☆ | ★★★☆☆ | ★★★★★ |
| Fairness | ★★★☆☆ | ★★★★★ | ★★★★★ |
Ethernet (LAN): Random access (CSMA/CD historically, now switched). Rationale: Bursty traffic, variable station count, low cost requirements. Collisions were acceptable trade-off for simplicity.
Wi-Fi: Random access (CSMA/CA). Rationale: Same as Ethernet, plus wireless constraints. Collision avoidance because detection is impossible.
Token Ring (historical LAN): Controlled access (token passing). Rationale: Deterministic access for time-sensitive applications, no collision overhead. Lost market share due to cost and complexity.
GSM (2G cellular): Channelization (TDMA + FDMA). Rationale: Continuous voice traffic, guaranteed quality, licensed spectrum.
3G cellular (CDMA2000, WCDMA): Channelization (CDMA). Rationale: Soft capacity, spectral efficiency, multipath resistance.
4G/5G cellular: Hybrid channelization (OFDMA with dynamic scheduling). Rationale: Combines channelization's guaranteed access with dynamic allocation for mixed traffic.
Satellite uplink: Channelization (TDMA/FDMA) or controlled access (polling/reservation). Rationale: Long propagation delays make collision recovery very costly.
Industrial control: Controlled access (polling, token, scheduled). Rationale: Deterministic timing is critical for real-time control loops.
Many real protocols combine elements from multiple categories. Wi-Fi's PCF (Point Coordination Function) adds controlled access periods to the normally random CSMA/CA. Reservation-ALOHA combines random access for control with scheduled data transmission. The categories are analytical frameworks, not rigid boxes.
Understanding the theoretical efficiency of each category helps predict real-world performance and guides protocol selection.
Throughput (S): The fraction of channel capacity used for successful data transmission.
S = Successful data bits / Total channel capacity
Ideal: S = 1.0 (100% efficiency)
Channel Utilization: Similar to throughput but may include control overhead as "useful."
Frame Delay (D): Time from when a frame becomes available until it's successfully transmitted.
Pure ALOHA:
S = G × e^(-2G)
Maximum: S_max = 1/(2e) ≈ 0.184 (18.4%) at G = 0.5
Where G = offered load (new + retransmitted frames per frame time)
Efficiency limited by "vulnerable period" of 2× frame time.
Slotted ALOHA:
S = G × e^(-G)
Maximum: S_max = 1/e ≈ 0.368 (36.8%) at G = 1.0
Slotting reduces vulnerable period to 1× frame time.
CSMA/CD (Ethernet): Efficiency depends on parameter α = propagation_delay / frame_time:
S ≈ 1 / (1 + 5α) [approximate for low α]
For typical Ethernet (small α), efficiency exceeds 90%.
Exact analysis involves complex contention slot modeling.
| Protocol/Method | Maximum Efficiency | Conditions |
|---|---|---|
| Pure ALOHA | 18.4% | G = 0.5 |
| Slotted ALOHA | 36.8% | G = 1.0 |
| Non-persistent CSMA | ~80% | Low propagation delay |
| 1-persistent CSMA | ~50% | Variable |
| CSMA/CD (10 Mbps Ethernet) | 95%+ | α < 0.05 |
| Controlled Access (Token/Poll) | Approaches 100% | High load, many stations |
| TDMA | 100% theoretical, ~90% practical | Guard time overhead |
| FDMA | 100% theoretical, ~85% practical | Guard band overhead |
Token Ring:
Efficiency ≈ 1 / (1 + a/N) [high load approximation]
Where:
At high load with many stations, efficiency approaches 100%. The token overhead becomes negligible compared to data frames.
Polling:
Efficiency = T_data / (T_poll + T_data)
Where:
Efficiency is high when stations have data (minimal polling waste) but degrades when many stations are empty.
TDMA/FDMA: Theoretical efficiency is 100%—each station uses its allocation completely. Practical efficiency is reduced by:
CDMA: Efficiency is more complex due to interference-limited capacity:
Capacity ≈ W/R × 1/(Eb/N0) × factor
Where W = bandwidth, R = data rate, Eb/N0 = energy-per-bit to noise ratio.
CDMA efficiency increases with more sophisticated receivers and power control.
Random access delay: Unbounded in theory (retransmissions can continue indefinitely). In practice, characterized by expected delay:
E[D] = T_frame × (1 + expected_retransmissions)
Expected retransmissions increase with load.
Controlled access delay: Bounded. Maximum delay = walk_time × N (time for token to visit all stations).
Channelization delay: Bounded. Delay = wait for next slot assignment (typically < frame period).
Real networks deviate from theoretical models. Error recovery, hardware limitations, protocol overhead, and implementation choices all reduce efficiency from theoretical maximums. Use these formulas for comparative analysis and approximation, not precise prediction.
We've examined the three major categories of MAC protocols comprehensively. Let's consolidate the essential understanding:
With the MAC protocol categories clear, we're prepared to explore the final foundational topic of this module: channel allocation. The next page examines the theoretical and practical frameworks for dividing channel capacity among users, including static vs. dynamic allocation, the tradeoffs involved, and how allocation decisions impact the protocols we'll study in subsequent modules.
Following this overview module, you'll dive deep into specific protocols: ALOHA, CSMA, CSMA/CD, CSMA/CA, token passing, and channelization techniques. The categorical framework from this page will help you understand where each protocol fits and why it was designed as it was.
You now understand the three fundamental approaches to MAC protocol design: random access, controlled access, and channelization. This conceptual framework will make every specific protocol immediately understandable—you'll recognize its category, predict its behavior, and understand its design tradeoffs.