Optimization of effective variance based CAC algorithms

The problem of connection admission control (CAC) in ATM networks is studied. The target is to find a realistic CAC algorithm implementation for real networks. The key requirements are that the scheme should be based on standardized traffic parameters, provide high utilization and the calculations should be sufficiently simple to be performed in real time. After comparing some promising schemes we discover that an effective variance based CAC algorithm produces very accurate results and thus concentrate on the determination of the model parameters. As the validity of the CAC scheme must be tested with diverse traffic cases a genetic algorithm based approach is used to optimize the parameter values