OPTIMAL ADMISSION AND ROUTING WITH CONGESTION-SENSITIVE CUSTOMER CLASSES

This paper considers optimal admission and routing control in multi-class service systems which customers can either receive quality regular is subject to congestion or congestion-free but less desirable at an alternative station, we call the self-service station. We formulate problem within Markov decision process framework focus on characterizing structure of dynamic policies maximize expected long-run rewards. For this, value function sample path arguments are used. The sensitivity modeled with class-independent holding costs results show how rewards customer classes affect their priorities stations. explore that priority for may not only depend also difference between have monotonicity properties, regarding decisions individual such they divide state space into three connected regions per class.