| With the rapid development of the Internet, the new types of the network application are gradually becoming rich, and the network scale is unceasingly increasing. As an effective solution to enhance the network control, the network traffic monitoring and predicting technique can not only obtain the network traffic data, but also have an important effect on the supervision management, service quality, safety management, fault detection and capacity planning of the network. To be a simple network management protocol, SNMP has been widely used in various network management systems. However, it produces a large number of management message because of polling MIB nodes, which has a high request to the bandwidth and the processing ability of the network, and it only supports centralized management. The RMON protocol, including two kinds of standards:RMON1 and RMON2, is able to solve these problems easily. Compared with RMON1, RMON2 can monitor more abundant traffic information periodically. This thesis implements the network traffic monitoring system based on the RMON2 protocol, and at the same time, the performances of two neural network traffic prediction algorithms are evaluated by computer simulation. The main work of this thesis is summarized as follows:1. For traffic monitor, the RMON protocol standard and working mode are introduced, with the discussion of the difference between RMON1 and RMON2. For traffic prediction, the BP neural network model and wavelet neural network model are investigated. And two kinds of neural network algorithms are analyzed and deduced in detail. The theoretical analysis demonstrates the wavelet neural network has better traffic prediction performance.2. The requirements of the RMON2 system are analyzed and its general design is put forward. Firstly, the sub modules, including the decomposition module of the zero layer, one layer and two layer are obtained by decomposing the whole system. Secondly, the design scheme of the whole system is achieved from the operation design processes of the one-layer and two-layer modules. Finally, the implantation of the RMON2 system is presented with C language.3. After establishing it, the RMON2 system is tested, in terms of the functional testing, performance testing, specification testing, combination testing, pressure testing and compatibility testing and so on. The testing results show that the system performance is stable.4. A kind of TCL script language is utilized to complete the RMON2 automation tools. Moreover, the method of monitoring network traffic for the RMON2 automation tools. And that the advantages and disadvantages of testing automation are analyzed in detail.5. The network operator’s equipment is monitored by the RMON2 automation tools. Interface uplink and downlink traffic values are sampled periodically, which will be used to the subsequent traffic prediction samples. Then the performances of the traffic prediction models of BP neural network and wavelet network are estimated through simulation experiments. Simulation results show that two kinds of prediction models can predict the traffic well and the network traffic prediction algorithm in the wavelet neural network obtains less prediction error under the same conditions. |
