| 摘要 |
<p>A method and system for automatically classifying in Virtual LANs network devices of a switched network. The classification can be performed by the network administrator from the network management workstation. The first step of the classification is to get network management information stored by the network devices concerning the activity on the network of the network devices; particularly this information can be the physical and logical addresses of the network devices in communication with the other network devices; for TCP/IP networks managed by SNMP protocol this information is stored in the MIB or the RMON matrix group variables of the network devices. Then, this input data feeds a neural network which may operate on the same network management workstation. More particularly, the neural network is a Self-Organizing Feature Map (SOFM) or an Adaptive Resonance Theory (ART) Artificial Neural Network. The output of the neural network is a list of device networks grouped in VLANs in such a way that the network devices communicating or having recently communicated together are grouped in a same VLAN. The input data is periodically updated in the network devices, consequently, the network device classification can be also periodically refreshed, the classification reflecting the real network device activity and thus be used to optimize the bandwidth in the switched network. <IMAGE></p> |
