Measurement and identification of EMI sources using pattern recognition and coding

The authors present a methodology for measuring and identifying EMI (electromagnetic interference) sources in complex electromagnetic environments, such as exist within digital computer systems. In particular, a MAP (maximum a posteriori probability) classifier with list decoding was applied to situations involving very low signal/noise ratios, commonly found in such a problem regime. Coding was applied to the test vectors to improve performance of the system. It was shown that very low error rates are possible with very noisy signals