Intelligent optical network traffic monitor design

The characteristics of a ground-to-ground radar link maintenance protocol for airfield communications are investigated. Such links are typically made at high speed over optical fibre in modern airfields and an appropriate method to identify both known traffic and unauthorised use is extremely important. Here, the pattern recognition technique of Support Vector Machines (SVMs) is employed for classification, recognising trends by supervised learning and using them to identify anomalies. The work is motive by the EU SESAR project which is developing a unified digital communication platform for air traffic management. To achieve this, it is necessary to gather network information and present it in segments. Packet transmission possesses two key parameters, packet payload and the time interval between transmissions. A simulation including these elements was performed, and SVM employed to search for traffic patterns. This was successful in that two known patterns were identified using a nonlinear polynomial SVM kernel. It is thus shown that it is possible to monitor discrete statistical traffic flows using SVM. Further payload types could be classified as anomalies enabling a significant reduction in the workload of the network engineer to monitor a critical optical network.