Loss systems in a random environment

TL;DR

This paper analyzes a single-server queue in a random environment, establishing conditions for product form steady states, exploring insensitivity properties, and examining the effects of environment-induced blocking and customer loss.

Contribution

It provides a necessary and sufficient condition for product form solutions in queue-environment systems and extends classical results to include environment interactions and loss phenomena.

Findings

Product form steady state characterized by an if-and-only-if condition.

Strong insensitivity property demonstrated for such systems.

Embedded Markov chain behavior differs from continuous-time process in these models.

Abstract

We consider a single server system with infinite waiting room in a random environment. The service system and the environment interact in both directions. Whenever the environment enters a prespecified subset of its state space the service process is completely blocked: Service is interrupted and newly arriving customers are lost. We prove an if-and-only-if-condition for a product form steady state distribution of the joint queueing-environment process. A consequence is a strong insensitivity property for such systems. We discuss several applications, e.g. from inventory theory and reliability theory, and show that our result extends and generalizes several theorems found in the literature, e.g. of queueing-inventory processes. We investigate further classical loss systems, where due to finite waiting room loss of customers occurs. In connection with loss of customers due to blocking by the environment and service interruptions new phenomena arise. We further investigate the embedded Markov chains at departure epochs and show that the behaviour of the embedded Markov chain is often considerably different from that of the continuous time Markov process. This is different from the behaviour of the standard M/G/1, where the steady state of the embedded Markov chain and the continuous time process coincide. For exponential queueing systems we show that there is a product form equilibrium of the embedded Markov chain under rather general conditions. For systems with non-exponential service times more restrictive constraints are needed, which we prove by a counter example where the environment represents an inventory attached to an M/D/1 queue. Such integrated queueing-inventory systems are dealt with in the literature previously, and are revisited here in detail.