Elementary queueing theory based on possibility theory (Q753661)

The paper considers a queueing system with multiple parallel servers, finite or infinite system capacity and calling source, but whose arrivals and service completions are governed by a possibility distribution. The system modelling and the changes in the number of customers therein is via the transition matrix and powers of this matrix. Powers of the matrix, under max-min composition oscillate or converge. If it converges, its limit produces the steady-state possibility distribution. An application is presented on the machine servicing problem to determine the optimal size of the repair crew. Finally, it remains an open problem as to what could be done with queueing decision models (like the machine servicing problem) when powers of the transition matrices do not converge and there is no steady-state possibility distribution.