Call admission control under QoS constraints using NeuroEvolution of topologies

In this paper, we solve the call admission control (CAC) problem in a cellular network via a form of NeuroEvolution algorithm. The results are compared with Complete Sharing scheme, which allows all types of traffic to access all available channels at all time, and show that our proposed CAC scheme CN can reduce the handoff failure subject to several predefined Quality of Service (QoS) constraints: the handoff dropping probability (CDR) and intra-class fairness are all guaranteed within the predefined constraints. Additionally, the numerical results shows that under the same handoff failure constraint, better fairness causes higher new setup calls blocking probabilities (CBR) and lower CDR, therefore the fairness and maximizing revenue are two opposite goals, and the fairness and CDR do not conflict with each other.