The Research Of QoS Evaluation Methods In Cognitive Network

In the background of the fast development of the network technology, the information on the internet is getting richer and richer, which is becoming the most important source in people's normal life and work. Since the network circumstance has been more and more complicated, there will be wrong with the network service more easily, which causes the performance of network to fall down sharply. In order to supply superior service for the users, it is obviously so essential to set up a perfect maintenance and management of the network. Studying the cognitive network, we can take in charge of the QoS management, make use of the characteristics of self-learning, self-management self-setting to improve the methods of QoS evolution.The computer network system consists of various terminals, and the topologies are different according to the different kinds of transmission medium, and each of them has its own access controlling method, so the evaluation of the network's performance is extremely complex. Currently, the three relatively mature methods are as follows:1)Test the computer network system and analyze all the parameters , so we can get the data and evaluate the performance;2) build a math model for the network system and work out the expressions to analyze the network system;3)Simulate the real network by running the computer programs, and then evaluate the network system's performance according to the results.The evaluation method used in this paper is based on the model, which collects the parameters, builds up the relationship between them, and analyses the data to evaluate the network's performance. This paper refers two evaluation models:Support Vector Machine(SVM) and Analytci Hierarchy Process (AHP) method . SVM model can improve the initialization, collect more suitable parameters and provide more accurate evaluation. At the same time, the AHP method combined with Fuzzy theory is discussed on the condition of the appropriate evaluation target, and the real-test data effectively proves the method's validity.