Design considerations of statistical multiplexors

Abstract

Design considerations for statistical multiplexors with Poisson message arrivals and the following three types of messages are considered: 1) constant length messages, 2) random length messages, and 3) segmented random length messages with fixed-size blocks. The operating costs of the multiplexors can be divided into transmission costs and storage costs. The transmission costs are dependent on the transmission rates of the lines and the number of lines used. The storage costs are dependent on the buffer length required, the cost of overhead in buffer management, and the cost of waste spaces of fixed-size block messages. The queuing delays imposed by multiplexors are dependent, for example, on delays due to buffering and computer scheduling algorithms. The trade-offs among transmission costs, storage costs, and delays are described. An optimal multiplexor is the one that yields minimum operating costs yet satisfies the required performances (e.g., delays, overflow probabilities). Some design considerations of multiplexing systems are then illustrated via examples.