LOW-POWER SPECTRUM ANALYSIS AND REAL-TIME DATA COMPRESSION SYSTEM

THE OBJECTIVE OF THIS PHASE I PROJECT IS TO CONSTRUCT AND DELIVER A "BRASSBOARD" FOURIER TRANSFORM SPECTRUM ANALYSIS PROCESSOR FOR MAGNETIC FIELD INSTRUMENTS HAVING A BANDWIDTH OF 0-10 HZ. IT WILL BE DESIGNED TO OPERATE AT A POWER CONSUMPTION OF LESS THAN 1 W AND INCORPORATES EXTENSIVE POST-PROCESSING DATA COMPRESSION CAPABILITIES. THIS CMOS BASED PROCESSOR CONSISTS OF A LOWPASS FILTER FOLLOWED BY AN A/D CONVERTER AND A DISCRETE FOURIER TRANSFORM (DFT) PROCESSOR, ALL UNDER THE CONTROL OF A SOFTWARE-DRIVEN MICROPROCESSOR WHICH ALSO PERFORMS POST-PROCESSING DATA COMPRESSION OPERATIONS. THE DFT PROCESSOR PRODUCES THE SAME RESULT AS AN FFT PROCESSOR BUT DOES NOT REQUIRE THE LARGE FFT INPUT MEMORY, SINCE ALL OPERATIONS ON A DATA SAMPLE ARE PERFORMED 'ON-THE-FLY" PRIOR TO INPUT OF THE NEXT A/D DATA SAMPLE. ALL MAJOR PROCESSING PARAMETERS INCLUDING FREQUENCY BAND PLACEMENT, RESOLUTION, AND THE LENGTH OF THE TIME AVERAGING WINDOW ARE REPROGRAMMABLE IN FLIGHT, AS ARE ALL POST-PROCESSING PARAMETERS AND ALGORITHMS. IN THE COURSE OF THIS PROJECT, AN INNOVATIVE POST-PROCESSING ALGORITHM MATCHED TO THE STATISTICS OF THE MAGNETIC FIELD DATA WILL BE DEVELOPED. POST PROCESSING CAPABILITIES WILL INCLUDE HANNING WEIGHTING IN THE FREQUENCY DOMAIN, A TWO-PASS NOISE-MEAN CALCULATION, ESTIMATION OF VARIANCE, ADAPTIVE THRESHOLDING AND THE DETERMINATION OF THE PEAK FREQUENCY, AMPLITUDE AND CLUSTER WIDTH OF SPECTRAL LINE CLUSTERS.