Main Article Content
In the multimedia communication networks providing quality of service (QoS), an interface between the signal processing systems and the communication systems is the call admission control (CAC) mechanism. Owing to the heterogeneous traffic produced by diverse signal processing systems in a multimedia communication network, the traditional CAC mechanism that used only one CAC algorithm can no longer satisfy the aim of QoS CAC: Utilize the network resource to the most best and still satisfy the QoS requirements of all connections. For satisfying the aim of QoS CAC in the multimedia communication networks, this study proposed an innovative CAC mechanism called black and white CAC (B&W CAC), which uses two CAC algorithms. One of them is called black CAC controller and is used for the traffic with specifications more uncertain, which is called black traffic here. The other is call white CAC controller and is for the traffic with clearer specifications, which is call white traffic. Because white traffic is simple, an equivalent bandwidth CAC is taken as the white CAC. On the other hand, a neural network CAC (NNCAC) is employed to be the black CAC to overcome the uncertainty of black traffic. Furthermore, owing to more parameters needed in a QoS CAC mechanism, a hierarchical NNCAC is proposed instead of the common used NNCAC. Besides to accommodate more parameters, a hierarchical NNCAC can keep the complexity low. The simulation results show the B&W CAC can obtain higher utilization and still meet the QoS requirements of traffic sources.