Self-learning security system for computer networks

Jul 09, 2009

Cyber attacks on computer networks are becoming increasingly commonplace. To counter the threat, they are protected by so-called network intrusion detection systems. But these fail to identify some attacks, or do not spot them until it is too late. To improve matters, Damiano Bolzoni of the University of Twente (The Netherlands) has developed a system which paves the way for a new generation of network security. This forms the subject of his doctorate, awarded by the Faculty of Electrical Engineering, Mathematics and Computer Science on 25 June.

A network intrusion detection system (NIDS) is like a kind of virus scanner, but for an entire network rather than a single computer. There are two types. The first draws upon a database of all known attacks, such as those attempted by . It works by recognizing the ‘signatures’ of methods previously used. But this means that it will not at first spot a new and as yet unknown method.

The second kind of NIDS uses anomaly detection. In other words, it learns how the is normally used and if it spots a deviation from this standard pattern it will alert the system administrator so that the suspected attack can be investigated. In practice, however, this type is not widely used because no really good systems are yet available commercially.

Bolzoni has been trying to change that by developing a new anomaly detection NIDS, which he has named SilentDefense. His system is based upon self-learning algorithms, which make it far more accurate than existing systems of this kind. Moreover, the chance of ‘false positive’ alerts is about 1000 times lower than in the systems currently available.

The system is now being further developed by SecurityMatters, the company recently founded by Bolzoni and fellow researchers Emmanuele Zambon and Sandro Etalle. They expect to launch SilentDefense commercially in mid-2010.

In Bolzoni’s view, the ideal NIDS is not of one type or the other but combines the two. For that to be possible, however, a good system based upon anomaly detection first needs to become available.

Provided by University of Twente (news : web)

Explore further: Researchers propose network-based evaluation tool to assess relief operations feasibility

add to favorites email to friend print save as pdf

Related Stories

The phony goat gets the worm

Mar 28, 2006

IBM researchers have designed a new way to detect and thwart attacks on computer networks. Code named "Billy Goat," the intrusion detection tool provides both early detection of worm attacks and fewer false alarms than other ...

New intrusion tolerance software fortifies server secrurity

Jun 16, 2008

In spite of increased focus and large investments in computer security, critical infrastructure systems remain vulnerable to attacks, says Arun Sood, professor of computer science at George Mason University. The increasing ...

Recommended for you

Large streams of data warn cars, banks and oil drillers

15 hours ago

Better warning systems that alert motorists to a collision, make banks aware of the risk of losses on bad customers, and tell oil companies about potential problems with new drilling. This is the aim of AMIDST, the EU project ...

User comments : 0

More news stories

Simplicity is key to co-operative robots

A way of making hundreds—or even thousands—of tiny robots cluster to carry out tasks without using any memory or processing power has been developed by engineers at the University of Sheffield, UK.

Microsoft CEO is driving data-culture mindset

(Phys.org) —Microsoft's future strategy: is all about leveraging data, from different sources, coming together using one cohesive Microsoft architecture. Microsoft CEO Satya Nadella on Tuesday, both in ...

Floating nuclear plants could ride out tsunamis

When an earthquake and tsunami struck the Fukushima Daiichi nuclear plant complex in 2011, neither the quake nor the inundation caused the ensuing contamination. Rather, it was the aftereffects—specifically, ...

Patent talk: Google sharpens contact lens vision

(Phys.org) —A report from Patent Bolt brings us one step closer to what Google may have in mind in developing smart contact lenses. According to the discussion Google is interested in the concept of contact ...