Intel is researching a new method of securing PCs and laptops by analyzing "normal" user behavior and raising a red flag when unusual activity occurs.
Researchers at Intel's Berkeley and Pittsburgh labs have been working on the project, dubbed Proteus for some time now, but they have not published details. However, in an interview published in Friday's edition of MIT's Technology Review magazine, Intel researcher Nina Taft discussed the project.
In a nutshell, Proteus is Intel's attempt to analyze user behavior on an individual device, just as enterprises have long done on networks. PCs equipped with Proteus will "get to know" a user's habits and behavior patterns, then raise an alarm when those patterns change, such as might occur when a laptop is stolen or infected by a botnet.
Proteus differs from current anomaly detection technologies because it can be personalized to a particular user, Intel says.
"We believe that today's solutions that configure security mechanisms for laptops are fundamentally flawed, in that all laptops are configured the same way," the Intel research team states in its Proteus project description. "Since anomaly detection relies upon finding outliers based upon some description of 'normal patterns of usage,' it is essential to define 'normal' correctly."
"Our premise is that normal behavior should be defined with respect to each end-host individually," Intel explains. "Since end-host behavior differs substantially across people, what is normal for one person may be out of range for another. We thus propose to develop and use individual profiles of end-host behavior that can, in turn, be used to allow the security solutions at each machine to be personalized. These profiles will also enable novel security solutions to be developed."
The research acknowledges the fact that end users engage in different behavior on different networks and locations, Taft says. For example, the technology would monitor the user's behavior in the office, on the road, and at home, identifying "normal" patterns of activity in each location. This will make it easier to recognize dangerous trends in activity while making it harder for attackers to hide their exploits, she says.
Taft tells Technology Review that Intel is currently testing Proteus internally with about 350 users and that she hopes to expand the test in the near future. No word yet on when the technology might be commercially available.
Have a comment on this story? Please click "Discuss" below. If you'd like to contact Dark Reading's editors directly, send us a message.