A new call admission control scheme based on pattern prediction for mobile wireless cellular networks

Recently, cellular networking has become really popular, because of its different and growing fields of interest. It is able to satisfy user requirements in terms of Quality of Service, especially when mobility is present, managing hand-over issues in an adequate manner. So, it is very important to decide whether a new connection can be accepted into the system, in order to maximize bandwidth utilization while avoiding quality degradations, with more emphasis for non-tolerant applications. In this paper, the attention is not focused on a particular prediction scheme, but it is shown how a statistical approach can enhance system performance. We employ a general predictor (that can be based on Markov theory, neural networks, data mining approach or similars) and then we integrate it with a threshold-based statistical bandwidth multiplexing scheme in order to propose the Prediction-Based MUltipleXing Call Admission Control (PB-MUX CAC) scheme for cellular networks with mobile hosts. It is able to manage users mobility and to mitigate it, minimizing the amount of wasted bandwidth. Many simulation campaigns have been carried-out, giving us the possibility of evaluating the performance of the proposed idea in terms of number of admitted flows, bandwidth utilization and call blocking/dropping probability.