The Dynamic QoS Research Of Next Generation Network Based On Traffic Prediction

With the rapid development of Next Generation Network (NGN) technology, integrated services (including traditional telecom services) can be provided on the basis of the IP bearer network. The independence and intercommunion scarcity between bearer layer and service layer make it hard to guarantee the QoS for NGN service, especially for the traditional real time communication service, which prevent the large-scale application of NGN. In view of this, the dissertation carries out a systematic research on the negotiation schemes between service admission control and IP QoS control and related problems.To comprehend the traffic properties of NGN services, the dissertation first analyzes the primary network traffic models. Then, the real collections of NGN traffic data are applied to analyze the important properties of NGN traffic enginerring. The analysis results indicate that the NGN traffic data features the periodicity, self-similarity and chaos.Based on the principles of wavelet signal separation and neural network signal forecasting, we apply the wavelet neural network method to predict NGN traffic and give an improved wavelet neural network prediction model based on the properties of wavelet neural network model and the real traffic forecasting requirements. The effects of the model are also evaluated in the dissertation.Based on the chaos properties of NGN traffic, the dissertation provides the chaos dynamic prediction model, which firstly rebuilds the phasic space of time sequence data, then computes its Lyapunov index and predicts traffic based on the parameter. The research results show that chaos based traffic prediction method can preferably fulfill the requirements for QoS dynamic control.Based on the IP QoS primary effects to NGN integrated service, the crucial IP QoS schemes and technologies are analyzed and summarized. In addition, the performance and the applied areas are analysed also.To build efficient negotiation scheme between NGN bearer layer and service layer, the NGN general QoS concept and its basic principles and models are proposed; the faults and limitations of the present NGN QoS methods static parameters estimating method and measure based dynamic resource configuration method are analyzed. Based on the previous work, the traffic prediction based NGN QoS scheme is given in the dissertation.On the basis of the analysis of characteristics of dynamic QoS realization, several key requirements of dynamic QoS scheme are summarized and two implementation schemes of dynamic QoS model, SNMP based scheme and mobile agent based scheme, are given. The performances of the two realization schemes are also analyzed.Finally, conclusions are made based on the dissertation work and the future work correspondent to the existent problems, are pointed out.