Ngn Bearer Network Performance Evaluation System Based On Neural Network Research

Currently, Next Generation Network NGN is a hot topic and focus in industry-wide communication. Communication network using the services delivery process, the majority of Internet users are focused on the problems, such as reliability and survivability of communication network, availability and controllability of network services. In order to understand the network of these features, network measurement technology gives some performance for the NGN bearing network parameters indicators, which include:connectivity, bandwidth, delay, packet loss rate, jitter and so on. But a single indicator is not conducive to reflect the overall characteristics of the network in the planning, construction or maintenance of the network. And, some indicators'content is too professional and goes against the users to understand the comprehensive network performance. People hope to construct a comprehensive evaluation model which have considered more performance indicators to understand network explicitly; The traditional network performance evaluation method need to build a membership function more or less. But it can't accurately describe the network level interval change characteristics and the design process of the human factors which are large, etc.Due to these problems, This dissertation presents a NGN bearing network performance evaluation system based on neural network, the idea is the most attention NGN bearing network selection of performance parameters, by using the neural network for pattern classification, solving nonlinear problem, and the characteristics of dealing mass of data, constructs a kind of network performance comprehensive evaluation model based on neural network. As it can convert several index to one which can reflect the comprehensive network performance.In order to facilitate comparative analysis, this dissertation selects four typical neural network models:BP network, radial basis function networks, self-organizing competitive network and feedback network. On the basis of the introduction-, of neural network theory, built a comprehensive evaluation model based on these four neural networks, through simulation experiments and case studies, comparing the characteristics of each model, and the four network output of the comprehensive evaluation index were analyzed comparison, verified based on NGN bearing network of comprehensive evaluation system correctness and practicality. And by improving comprehensive evaluation system,we constructs network performance prediction model based on BP neural network and radial basis function network.By the analysis of example and prove the neural network NGN network performance prediction bearing network the validity of the model and the reference value.