| Summary: | This work studies the adaptive scheduling in the QoS supported networks. It analyses the existent scheduling disciplines and chooses several basic ones. For the chosen disciplines, the adaptive models are presented that are capable of finding the optimal scheduler configuration based on the QoS requirements and a provider’s preferences concerning each service class. The preferences are expressed with the prices for network services, however, the presented models are flexible in that other criteria can be exploited. It is shown that the proposed adaptive models can be integrated easily into the existent QoS frameworks, such as IntServ and DiffServ. Being independent of a particular signalling protocol and an underlying transport protocol, the models can also be deployed to other QoS environments. The theoretical correctness of the proposed models is confirmed by the extensive simulations carried out with the NS-2 simulator. Each adaptive model is tested within the IntServ and DiffServ QoS frameworks with various requirements and number of service classes. The final simulation scenario presents a hybrid framework, in which the signalling mechanisms from the IntServ QoS framework are combined with the forwarding mechanisms of the DiffServ framework.
|
|---|