Why CSMA/CA and not CSMA/CD in 802.11?

Posted on by

Carrier Sense Media Access (CSMA) is a network access mechanism that has been around since the early days of networking. CSMA was developed to overcome the operational issues found in ALOHAnet, the first wireless packet data network in the early 70s.

ALOHAnet consisted of a hub-and-spoke wireless network where the spokes (located on several Hawaiian islands) communicated with a hub. All the radios operated on the same pair of uplink/downlink frequencies, and each station transmitted randomly – without checking whether other stations were transmitting at the same time.

Upon each transmission, the sending station expected an ACK packet. If this ACK didn’t arrive, it assumed that the transmission had failed and had to be re sent.

As you can imagine, this was highly inefficient, even with the limited number of spokes in the network. A “listen before sending” protocol was required, which led to the development of CSMA – and its variations.

CSMA/CD

The CSMA/CD (Collision Detection) variation made its way to the Ethernet protocol. In Ethernet, a station begins to transmit a frame when the media is idle (the “listen before sending” part), and keeps listening to the media for collisions.

If a collision occurs, the transmitting station stops sending (and jams the media for a short while to make sure that every other station is aware of the collision). After a set time, the transmission attempt begins all over again.

CSMA/CA

The CSMA/CA variation was adopted by the 802.11 WiFi protocols. In a nutshell, the “listen before sending” part is the same as in CSMA/CD, but in this case, the transmitting station doesn’t keep listening to the media for collisions. It transmits its full frame, unaware of whether a collision has occurred or not.

Instead, the sending station expects an ACK packet from the receiver (same as in ALOHAnet). If it doesn’t come, then it assumes that a collision did occur, and the transmission attempt begins all over again – after some backoff time.

Why not CSMA/CD in 802.11 ?

It may have seemed more efficient to adopt CSMA/CD for wireless communications – after all, in CSMA/CD the sending station stops its transmission as soon as a collision is detected, while in CSMA/CA, the sending station sends its full frame, totally oblivious to the media’s state. So why was CSMA/CA adopted instead?

It all boils down to the media sense. In 802.11, there is no assurance that the sending station can hear the frame sent by another station, which can be too far away for its signal to be decoded as 802.11 energy. Since the collision cannot be detected (CD), the only other option is to try avoiding the collision (CA) altogether.

On the other hand, CD is suitable for Ethernet, since there is assurance that all the stations in an Ethernet segment can hear each other. This made more sense in the era of Ethernet hubs though, which have been replaced by Ethernet switches. With switches, there are only two stations on each Ethernet segment  – the station and the switch itself, making collisions more unlikely, and totally impossible in full-duplex.