Based On The Design Of Distributed Industrial Ethernet Real-time Detection System Of Self-similar Model

With the Ethernet and the TCP/IP entering the control field, Ethernet is able to be equipped in the real time industrial field network. Developed countries and the field bus organizations have paid a lot for importing the Ethernet according to their own ability, and have proposed the industrial control system plan based on Ethernet. In this field, our country has also placed more emphasis on engineering application of the industrial Ethernet control network.The commercial Ethernet may not be influenced deeply by a little error packages, but a normal control data error may cause a serious accident in the control system. As a result, more attentions should be paid to studying the Ethernet faults and diagnosing in time so as to improve reliability and robust of whole network.The network traffic characteristic shows the working status of the industrial Ethernet accurately, and abnormal traffic will also influence the network running performance largely. The overheavy network traffic would bring frequent package collisions and network delay, so detecting the exception of the Ethernet traffic will guarantee the industrial Ethernet to work reliably.The qeuing model of Ethernet traffic used widely in the conventional Network Traffic Detecting System (NTDS) cannot describe the network traffic characteristic perfectly. The Network Traffic Detecting System used traditional queuing model of Ethernt traffic is not real-time, and cannot work perfectly at the situation of complex topology.This thesis has designed a novel Network Traffic Detecting System based on this background, which adopts the self-similar model to detect Ethernet traffic characteristic. The complexity of the original algorithm to calculate Hurst parameter has been reduced from linear to constant. The NTDS described in the thesis adopts the distributed framework, which is implemented with C/S structure. The Linux Clients capture the raw data from their subnets, and then transmit analytical data and alarm information to the Windows Server. The Windows Server dispalys the information aggregated from all the clients, and also we can set the parameters in the LinuxClients with Windows Server interface. The results of strict tests reveal that the NTDS proposed in this thesis has some characters such as real time, time saving and high precision. The Network Traffic Detecting System has been used in Hangzhou iron and steel works to monitor its Ethernet state.