An Improved Algorithm of the Optimization Control Problem of Loss Systems

The optimization control problem of GI/M/n loss systems could be modeled with Markov decision processes (MDP); however, there have been no effective algorithms to solve the optimal policies of such large-scale problems by now. The model characteristic of the optimization control problem of GI/M/n loss systems is well analyzed, and the improved algorithm to solve optimal policies of the problem is proposed correspondingly. The algorithm is mainly improved in the selection rule of initial policy, the improvement rule of policy and the evaluation criterion of optimal policies, so both the storage space and computing time are reduced. Meanwhile the optimal solution of the MDP problem could be obtained by the improved algorithm. Finally, a simple comparison between the improved algorithm and conventional algorithm is given through an example. It can be concluded that the improvement algorithm is suitable to solve the large-scale optimization control problem of GI/M/n loss systems