A Learning Approach for Call Admission Control under QoS Constraints in Cellular Networks

A new approach for learning call admission control (CAC) schemes that can provide quality of service (QoS) guarantees to ongoing calls from several classes of traffic with different resource requirements is presented. Comparison with two other CAC schemes shows that the new approach is not only capable of utilizing the network resource to maximize revenue but also maintain the handover dropping rate (CDR) under a prescribed upper bound while still maintaining an acceptable call blocking rate (CBR). The learning CAC scheme is shown to work successfully in the presence of smoothly changing arrival rates of traffic. The CAC policy is obtained through a form of neuroevolution (NE) algorithm