Traffic Control of Two Parallel Stations Using the Optimal Dynamic Assignment Policy

In this chapter, the authors study some optimal dynamic policies of assignment to control the entrance of a system. This system is formed of two waiting lines (queues) in parallel. Every line is composed of a server and a waiting room (buffer). Upon arrival to this system, the authors suppose the existence of an assignment policy. The role of the policy is to assign dynamically the customers (the customers here may represent a physical customer in the entrance of a cinema or bank, packets in a computer networks, calls in a telephone system…etc.) to the one or to the other of two lines or it may reject (or assigned to other system). This assignment is done according to policies in order to achieve some performance criterion.

Parallel fluid queues with constant inflows and simultaneous random reductions

We consider I fluid queues in parallel. Each fluid queue has a deterministic inflow with a constant rate. At a random instant subject to a Poisson process, random amounts of fluids are simultaneously reduced. The requested amounts for the reduction are subject to a general I-dimensional distribution. The queues with inventories that are smaller than the requests are emptied. Stochastic upper bounds are considered for the stationary distribution of the joint buffer contents. Our major interest is in finding exponential product-form bounds, which turn out to have the appropriate decay rates with respect to certain linear combinations of buffer contents.

Parallel fluid queues with constant inflows and simultaneous random reductions

We consider I fluid queues in parallel. Each fluid queue has a deterministic inflow with a constant rate. At a random instant subject to a Poisson process, random amounts of fluids are simultaneously reduced. The requested amounts for the reduction are subject to a general I-dimensional distribution. The queues with inventories that are smaller than the requests are emptied. Stochastic upper bounds are considered for the stationary distribution of the joint buffer contents. Our major interest is in finding exponential product-form bounds, which turn out to have the appropriate decay rates with respect to certain linear combinations of buffer contents.

The Downs-Thomson Effect in a Markov Process

Suppose customers pass through a network of two queues in parallel. A statedependent routing policy gives individuals their quickest journey. The Downs-Thomson effect is any increase in the long-run expected journey time caused by an increase in the service rates. This effect may occur.

A note on waiting times in systems with queues in parallel

Numerical data are presented concerning the mean and the standard deviation of the waiting-time distribution for multiserver systems with queues in parallel, in which customers choose one of the shortest queues upon arrival. Moreover, a new numerical method is outlined for calculating state probabilities and moments of queue-length distributions. This method is based on power series expansions and recursion. It is applicable to many systems with more than one waiting line.

A note on waiting times in systems with queues in parallel

Numerical data are presented concerning the mean and the standard deviation of the waiting-time distribution for multiserver systems with queues in parallel, in which customers choose one of the shortest queues upon arrival. Moreover, a new numerical method is outlined for calculating state probabilities and moments of queue-length distributions. This method is based on power series expansions and recursion. It is applicable to many systems with more than one waiting line.

The shortest queue problem

A Poisson stream of customers arrives at a service center which consists of two single-server queues in parallel. The service times of the customers are exponentially distributed, and both servers serve at the same rate. Arriving customers join the shortest of the two queues, with ties broken in any plausible manner. No jockeying between the queues is allowed. Employing linear programming techniques, we calculate bounds for the probability distribution of the number of customers in the system, and its expected value in equilibrium. The bounds are asymptotically tight in heavy traffic.