Another popular form of LAN is the token-ring in which each computer of the LAN is connected to two neighbors so that they all form a big ring. Whenever one wants to send a bit stream to another, it alters the voltage on the wire to its downwind neighbor, which then decodes the packet's address. If it is the intended recipient, it consumes the packet. Otherwise, it retransmits it to its downwind neighbor, and the process repeats until the packet makes it to the intended recipient.

Of course, two or more computers might try to send at the same time, so some method must be used to arbitrate. The ancient method of passing a "talking stick" around from person to person is used, only the stick is just a special sequence of 1's and 0's, a special pattern called a token. A computer in a token ring can only transmit if it gets the token next. Otherwise, it must wait until a token comes around.

Fig. 23.10.1 shows a drastically simplified picture of a token ring LAN. We do not have further space to explore the intricacies and problems of token rings herein. One of the most serious of these problems relates to when the token is destroyed or lost, or one of the computer dies in the ring. Of course, these problems have all been solved, to at least some satisfactory degree, in commercial token-ring products, which are quite popular and perform well.