Dynamic optimal policy of feedback queueing systems with multi-type customers
In this paper,we study the optimal admission control and dynamic pricing policies for feedback queueing systems with multiple customer types and finite buffers.Assume that customers arrival system is a Poisson flow,the service time of customers follows an exponential distribution,and the customer types are determined by different net profit values.Under the objective of maximizing average net profit,from the perspective of policy structure,we prove the existence of the optimal threshold policy of admission control and the existence of the optimal monotone price policy,give a sufficient and necessary condition and some related properties for the system to be profitable in the case of single server and no waiting cost.In addition,an effective backward iterative algorithm is proposed based on the special structure of the model.Finally,we use several numerical experiments to verify the results,observe the difference of the optimal profit value of different regulation methods,and compare the computational efficiency with that of the classical policy iterative algorithm.
NETL
NSTL
万方数据
