In a local area network, each node is capable of transmitting over the same link cable. The set of rules governing access, usage time and monitoring constitutes the access protocol for the cables or communication media.

There are three main protocols for controlling access to the medium:

· Contention CSMA (Carrier Sense Multiple Access)

· Passing the token (Token)

· TDMA (Time Division Multiple Access)

1. CSMA/CD — Collision Detection Multiple Access

In modern switched, full-duplex Ethernet networks, CSMA/CD is no longer used in practice, but remains important for understanding how Ethernet used to work

In a MAC-layer contention protocol, each node has equal access to the medium. A signal detection system is used to identify a signal on the medium. When a node has a frame to transmit, it scans the medium to determine whether it is occupied by another station. If the medium is free, it may transmit.

Several nodes may detect that the medium is free and start transmitting immediately. If two or more nodes start transmitting at the same time, a collision occurs. When a collision is detected, the nodes sending the messages must retransmit them. Each node must wait for a random delay before attempting to retransmit the messages, which reduces the likelihood of another collision.

CSMA/CD (CDfor Collision Detection) detects collisions when two stations attempt to transmit at the same time. Once a collision is detected, the system calculates a random waiting time (exponential backoff) for each station. The station with the shortest waiting time will transmit again first.

The CSMA/CD protocol sends a broadcast transmission to all stations. All stations on the network listen to the medium and accept the message contained in this broadcast frame. Each message has a destination address. Only the workstation with an address matching the message’s destination address will interpret the message’s content.

CSMA/CD is a fast and reliable method because, under normal conditions (without excessive load or hardware issues), there are few collisions. Despite the collision detection mechanism, some collisions may go undetected in non-standard configurations.

📄 The problem of short frames and collisions

If stations A and B are far apart on the network, A may send a very short frame, listen for its echo, and assume that everything is fine. However, it is possible that B is listening on the other end, that A’s frame has not yet arrived, and so B sends its own frame. A collision will then occur even though A believed everything had gone smoothly — A’s frame would be lost.

 

To prevent this, the standard requires a minimum frame size of 512 bits. If the message is not long enough, padding bits are added to reach this size. However, this is not sufficient: if the network size is not limited, the problem can still occur.

The size of the network is therefore limited based on the round-trip delay (Round Trip Delay) of the minimum frame and the data rate. In other words, it depends on the time it takes for 512 bits to travel from one end of the network to the other and back. To detect a collision, the signal from the first bit must have reached the far end before the station has finished transmitting its 512 bits, and if a station at the far end has transmitted a bit at that moment, it must have had time to arrive. In summary: the transmission time for 512 bits must be greater than the network’s Round Trip Delay.

⚡ CSMA/CA — collision-free variant

CSMA/CA (CAfor Collision Avoidance) aims to prevent collisions rather than detect them. This protocol is used in IEEE 802.11 wireless networks (Wi-Fi) where collision detection is not possible via radio. It relies on the RTS/CTS (Request To Send / Clear To Send) mechanism and a random delay (DIFS + backoff) before each transmission to minimise the risk of collisions.

2. OFDMA and MU-MIMO (mechanism, principle)

Wi-Fi 6 (802.11ax) introduces two complementary mechanisms that radically transform the management of the radio channel. OFDMA (Orthogonal Frequency Division Multiple Access) divides each channel into sub-carriers known as RUs (Resource Units), allowing an access point to serve multiple devices simultaneously on separate frequency bands, without each client having to wait its turn. This represents a fundamental departure from previous generations, where the channel was allocated to a single transmitter at a time.

3. OFDMA and MU-MIMO (complementarity and practical benefits)

MU-MIMO (Multi-User Multiple Input Multiple Output) operates on a different level: it uses multiple antennas to transmit independent data streams to multiple clients in parallel, both during transmission (downlink) and reception (uplink) via Wi-Fi 6. When combined, OFDMA and MU-MIMO enable an access point to efficiently manage high-density environments — open-plan offices, lecture theatres, hotels — where previous generations quickly became overwhelmed by contention.

4. Passing the token

This protocol comes in two forms:

· The token-passing ring is used in ring topologies.

· Logical token passing (logical token passing) is used primarily in a technology known as Arcnet.

The token passing technique is the second medium access control protocol. This protocol is used in bus and ring topologies. Each node has an equal chance to transmit. The right to transmit is granted by the token, which is passed sequentially from one node to the next. Only the node holding the token may transmit a message.

How it works :

· Wait for the transmission token to arrive. The token circulates and passes from node to node in sequence.

· If the token is received and there are no messages to send, forward the token to the next node.

· If the token is received and there is a message to be sent, only the token holder may transmit it. The message is picked up en route by the recipient, who sends an acknowledgement of receipt back to the sender.

· Once the message has circulated around the ring, it is retrieved by the sender, who checks that it has been received correctly before destroying it and releasing the token.

· The token has been passed to the next node.

With the ring token, the token follows the physical order of the stations, whereas with the circulating token, it follows the logical number found on each station’s network interface card.

The token-passing method is very reliable because only one node can transmit at any given time, so collisions are impossible. As all nodes have regular access to the cable, each one is served equally. However, this technique introduces a delay compared to the CSMA/CD contention method.

⚠ Token passing — obsolete in LANs

Token Ring (IEEE 802.5) and Arcnet are now completely obsolete in corporate local area networks. These technologies were superseded by switched Ethernet from the 1990s onwards. The token ring mechanism is still used in certain industrial protocols (such as PROFIBUS) and in field networks.

 

5. TDMA (Time Division Multiple Access)

In this method, time is divided into time slots allocated to each node. Thus, a station may transmit a message during one or more of the time slots allocated to it. At other times, it waits its turn to transmit. A priority station may, by configuration, be allocated more time slots than another station. This method prevents collisions.

TDMA is rarely used in LANs today. However, it is widely used in cellular wireless networks (GSM, 2G) and in satellite communication systems, as well as in OFDMA/TDMA in Wi‑Fi 6/7 (which remains a form of time/frequency division multiple access).

Modifié le: jeudi 8 octobre 2026, 12:26