A New Algorithm for The Design of Finite Capacity Service Units
Dario Andrea Bini, Srinivas Chakravarthy and Beatrice Meini

Abstract. In this paper we present a new algorithm for the design of finite capacity service units with phase type arrivals and phase type services. These systems are very useful in manufacturing and production areas. Using cyclic reduction method that exploits both inner and outer structures of the generator of the Markov process governing PH/PH/1/$N$ queue, we develop a new, fast and numerically efficient algorithm for solving the steady state probability vector. This fast algorithm can interactively be used in solving a number of optimization problems that require searching for an optimal value of $N$.  Comparison of our algorithm to that of block Gauss-Seidel iterative procedure and to the matrix sign function approach for selected examples is presented..

 <dvi>  <postscript>  files of the paper