Statistical Multiplexing

Why
Statistical Multiplexing is important because it allows a packet switched network to support more users.

What
Statistical Multiplexing is a multiplexing strategy in which we simply send packets into a network without reserving resources in advance. It relies on one key assumption:
 * Not all traffic comes in at one burst

Example 1
Assume that we have: When we have less than 11 users, then our link can support the data that we're putting through it. If we have 11 or more users, then our performance starts to dip under 100kbps due to queuing delay. However, we note that this occurs with probability .0004, which is almost never given users are doing random accesses (not exactly a safe assumption)
 * A 1Mbps link
 * 35 users, each of which are with probability .1 and generates 100kbps of data when they are active

Example 2
Assume we have: Under statistical multiplexing, this user can transfer his data at the full link capacity of 1Mbps. However, if we were to use time-division multiplexing with 10 divisions of 100kbps capacity each. It would take 10 seconds using the TDM while it would only take 1 second with statistical multiplexing.
 * A 1Mbps link
 * 35 users, one of which is active and 34 which are inactive
 * The active user generates 1Mbps of data