|The U.S. Navy seeks a partner for licensing a test system combining a sensor array with a machine learning system to evaluate the probability a microelectronic device is counterfeit.
It is increasingly important to have the ability to detect defective or counterfeit microelectronic devices in the marketplace. Defects or counterfeit devices can include parts that do not conform to their specification, are not authorized by an original equipment manufacturer, are being passed off as a new part, or have been subjected to one or more damage or stress events exceeding acceptable limits such as an electrostatic discharge event. System defect or supply chain problems are becoming more difficult given large volumes, difficulty in accessing parts in an assembly, and different sizes, shapes, and structures for mass produced parts. Thus, there is a need to improve electronic system supply chain defect detection capabilities which can be used at any stage in a supply chain.
NSWC Crane has developed and patented a test system that combines an array of sensors with a machine learning system. A common problem with existing methods of acquisition and comparison of parts in a supply chain is that they are generally not good at accounting for normal manufacturing process variations, which can vary with device lots, foundries, etc. The test system uses multiple detection and data collection methods coupled with one or more decision engines such s neural networks, image recognition, statistical correlation tools, and decision trees which can incorporate various learning processes. This combination of sensors and adaptive machine learning tools has the ability to detect counterfeit devices with more accuracy by taking into account normal variations between devices.
Apr 26, 2016
Purdue Office of Technology Commercialization
The Convergence Center
101 Foundry Drive, Suite 2500
West Lafayette, IN 47906
Phone: (765) 588-3475
Fax: (765) 463-3486