CMOS APS Image Sensors For Compact Aerial Vehicles

The Prediction Systems, Inc./New Jersey Institute of Technology (PSI/NJIT) team hassuccessfully demonstrated the applicability of Artificial Intelligence techniques,including Neural Networks (NN) and other techniques, in solving the intrusiondetection problems. Sufficient components of the Hierarchical Intrusion DetectionEngine (HIDE) were built in the Phase I effort to demonstrate that our adaptiveapproach can effectively detect a flooding attack in computer networks. Our rate offalse positives and false negatives were less than 0.4% when attempting to detectthe attack using a 4-second window. The rates of false positives and false negativesdrop even lower when we increased that window to 16 seconds. Our success can beattributed to the use of probability density functions (PDFs) (versus averages,variances, etc.), the compression of PDFs, and the efficient use of innovative typeof NNs. This tool, if deployed in the Army's Tactical Internet, can remove therequirement of having high personnel expertise needed to examine the data (e.g.,System logs) to detect intrusions. During the Phase II effort a refinement of thecomponents built in Phase I plus additional components (for added flexibility, easeof use, and more intrusion detection capabilities) and integration of the HIDE toolare being proposed.