Multiplexers also are used in building digital semiconductors such as central processing units (CPUs) and graphics controllers. In these applications, the number of inputs is generally a multiple of two, the number of outputs is either one or relatively small multiple of two, and the number of control signals is related to the combined number of inputs and outputs. For example, a two-input, one-output multiplexer requires only one control signal to select the input, and a 16-input, four-output multiplexer requires four control signals to select the input and two to select the output.
Types of multiplexers also are used in communications. A telephone network is an example of a very large virtual multiplexer that is built from many smaller, discrete ones. Instead of having a direct connection from every telephone to every other telephone — which would be physically impossible — the network muxes individual telephone lines onto a small number of wires as calls are placed. At the receiving end, a demultiplexer, or demux, chooses the correct destination from the many possible destinations by applying the same principle in reverse.
There are more complex forms of multiplexers. Time-division multiplexers, for example, have the same input/output characteristics as other multiplexers, but instead of having control signals, they alternate between all possible inputs at precise time intervals. By taking turns in this manner, many inputs can share one output. This technique is commonly used on long-distance phone lines, allowing many individual phone calls to be spliced together without affecting the speed or quality of any individual call. Time-division multiplexers generally are built as semiconductor devices, or chips, but they also can be built as optical devices for fiber optic applications.
Even more complex are code-division multiplexers. Using mathematical techniques developed during World War II for cryptographic purposes, they have since found application in modern code division multiple access (CDMA) cellular networks. These semiconductor devices work by assigning each input a unique complex mathematical code. Each input applies its code to the signal that it receives, and all signals are simultaneously sent to the output. At the receiving end, a demux performs the inverse mathematical operation to extract the original signals.
