A representative validation of a neural 3G admission control through rules extraction

The proposal of new mechanisms and systems of call admission control has as objective the ensuring quality of service (QoS) for mobile networks. These mechanisms are evaluated through simulators, presenting some computational limitations. Recently, CAC-RD (Call Admission Control based on Reservation and Diagnosis) has been proposed as a new mechanism to simulate 3G UMTS networks, but a computational resources limitation was found. As example, to simulate 600 s of traffic with a limit of 1100 users, the simulation spent approximately 19 hours. This prevents a more real simulation in the presence of greater number of users. Recently, to surmount these limitations, the authors have presented CAC-RDNN (CAC-RD Neural Network), a new approach based on neural networks to represent quantitatively the CAC-RD, but it was verified that the new representation fails when qualitative aspects are evaluated. In this paper, the qualitative analysis of the mathematical representation of CAC-RDNN is evaluated. Results showed that CAC-RDNN reflects CAC-RD modules behavior in both quantitative and qualitative aspects, being able to simulate real networks.