Recognition and Identification of Anomalies in Web Traffic Increasingly
Complex. The leader in secure web enabled delivery solutions, Bee Ware, has embarked
on a research program in conjunction with top French datamining research labs, the LIRMM and the LGI2P. Their common goal is to improve traffic analysis in response to the new challenges introduced by the web.
“The program is very much part of our next generation ICX engine development,” said Eric Battistoni, Bee Ware Marketing Director. “ICX correlates various analytic algorithms going from bot detection to session tracking. The research explores new avenues for identifying traffic and aims to develop more precise analytic methods that are also easier to run.”
Web applications have become an essential part of the life of an organisation, and their use extends to both professional and personal daily business. However, along with dramatic growth there is a considerable increase in web traffic diversity. Nascent hypermedia interfaces, the Web 2 model and XML provide for richer content but also unleash floods of unfiltered data. Recognizing and identifying traffic anomalies in this context is a very real challenge, firstly from a human perspective but even more so from a technical one. Regular incident detection mechanisms are simply not capable of differentiating between false alarms and unidentified attacks.
“Exploding web traffic illustrates the new complexity of data analysis very well. New approaches for faster and more accurate analysis are required to confront the quantity and diversity of the data we are looking at” explained Maguelonne Teisseire, Lecturer in Polytech’ Montpellier and Head of the TATOO project at the LIRMM.
Pascal Poncelet, Professor and Head of the KDD project (LGI2P), confirmed
the thought trend: “New approaches such as behavioral modeling or the search for time sequences make it possible to extract relevant information and support decision-making. This has strong potential value when it comes to computer security.”
The research program will last 18 months and be punctuated by demonstrations of results obtained. It will end with the integration of selected algorithms into the Bee Ware product range.
Kaynak : 