Performance of two different soft-computing based controllers for service selection and price adjusting tasks of the ISPs of the DS enabled Internet

In the differentiated services (DS) enabled Internet, service selections and price settings for service level agreements are important tasks for the Internet service providers (ISPs). These operations affect directly the economic profit of the ISPs as well as quality of end-to-end transmission services. Implementation of a controller for the service selection and price setting tasks is challenging, because the control system must be able to operate without exact information about the controlled environment. In addition to this, the control system should dynamically adapt to agreement establishing strategies of different customers. In this paper, we study the usability of two different soft-computing based control systems for the control tasks. The first control system is based on the use of simple manually tuned Mamdani type of fuzzy controllers, while the second system uses the self-organized feature map and genetic algorithms methods for implementing automatically tuning controller for the control tasks. The fuzzy controller based control system is designed so that a minimum amount of information is needed for control actions, while the automatically tuning controller uses several input variables. Performance of the alternative controllers is tested.