Sharing a Common Channel : Media Access Protocols

In this course so far, we have studied two kinds of links that connect computers together: a point-to-point link, typically constructed from a wire, and a shared medium such as radio, for which we developed frequency division multiplexing. In a point-to-point link, the sharing of the link can be done by the switch to which the link is connected. In contrast, when using a shared medium to connect end points together, we might not be able to solve the problem by having all the sharing logic implemented by a single switch—indeed, there may be no switch at all, if the end points are all peers of each other. More pertinently, even if there were a switch present, the information about the queues on the different nodes is distributed, and the overhead of a centralized switch deciding how the medium must be shared must be considered and minimized. This lecture discusses some ways of sharing a common communication medium, or channel, amongst multiple nodes. We start by giving some examples of shared media. Then, we outline three ways in which one can share the medium: frequency division multiplexing, time division multiplexing, and contention protocols. Each of these ways is actually a family of protocols, and each must provide a way to handle state distributed across the nodes. Most of this lecture will be about contention protocols, which provide a fully distributed solution to the problem, and are used widely in many systems today.