Intelligent agents for proactive fault detection

As the Internet becomes a critical component of our society, a key challenge is to maintain network availability and reliability. Intelligent processing agents that reside at network nodes use an adaptive learning method to detect abnormal network behavior before a fault actually occurs. In a test at Rensselaer Polytechnic Institute, an agent on a router detected a file server failure 12 minutes before it occurred.