Qos Parameter Prediction And Optimization Technology In IP Network

With the continuous development of IP network technology, IP network is transforming from a single data-transmitted network to an advanced network, which can not only transmit data, voice, image but also multimedia. In the meantime, the expectation of QoS (Quality of Service) for users is becoming higher and higher. So in order to guarantee the end-to-end QoS goal and satisfy user’s demands, it is quite imperative to improve the quality of QoS prediction and optimization. When QoS does not meet the needs of users, IP network is required to take some strategies and methods to improve the overall quality of network according to the predicted QoS parameters and user demands.This paper mainly deals with the architecture of QoS parameter prediction and optimization and its realization method in IP network. The architecture includes six modules:network information collection module, QoS parameter prediction module, user-needs extraction module, database module, case-based reasoning module and QoS optimization execution module. Among them, we mainly introduce QoS parameter prediction module and case-based reasoning module. In the QoS parameter prediction module, in consideration that RBF neural network has many advantages including its self-adaptive structure, nothing between output and initial weight value, fast learning speed, less time to train and better accuracy, this paper puts forward a RBF neural network QoS parameter prediction method, combining the K-means clustering algorithm and least squares algorithm to get base function parameters and the network weights, eventually realizing the QoS prediction. Besides, simulation experiment select a group of QoS historical data to predict the future QoS, and the experimental results demonstrate that the forecasting result and the actual result has higher degree of fitting. In the case-based reasoning module, this paper puts forward a kind of IP network QoS optimization scheme method based on case-based reasoning technology (CBR). This method combines QoS parameter predicting results and specific needs of users to generate a target case, and then retrieve the optimal case and utilize its solution to solve the QoS optimization problem. In addition, case modification and case study methods are adopted to match the different situations of QoS optimization problem. Finally, simulation experiment provides a rerouting scenario to demonstrate the feasibility and efficiency of the optimization scheme.