Abstract
In this paper, we explore the analogy between queueing theory and information theory, especially in traffic regulation and source coding. We consider a traffic regulation problem, where an input sequence is mapped to an output sequence. The rate of a regulator is defined to be the 'peak' rate of the output sequence and the performance of a regulator is in terms of loss probability. Under our formulation, we show theorems that are analogous to Shannon's source coding theorem and the universal block coding theorem. For real-time traffic regulators that have a maximum delay constraint, the trade-off between delay and loss probability is characterized by the recently developed notion of effective bandwidth.