COMPUTER ARCHITECTURE FOR KALMAN FILTERING

SIGNAL PROCESSING ALGORITHMS ARE MULTIPLY/ACCUMULATE INTENSIVE. NOVEL METHODS USING ADVANCED IC DEVICES ARE NEEDED. RADAR DIGITAL SIGNAL PROCESSING AND SPREAD SPECTRUM PROCESSING TASK INCLUDE FFT, LMS, L, CONVOLUTION, AND SPECTRUM ANALYSIS. AS EXAMPLES, 32 X 32 MULTIPLIER, 64 X 64 MULTIPLIER, PARALLEL MULTIPLIER AND FLOATINGPOINT PROCESSING ARCHITECTURES ARE SOUGHT FOR PROCESSING SPEEDS AT 10 NANOSECOND. MICROPROGRAMMABLE DESIGNS ARE TO BE STUDIED FOR DEVELOPING HIGH SPEED ARCHIETECTURES. EXPLORATORY DEVELOPMENT IS PROPOSED TO DESIGN AND DEVELOP SPECIALIZED HIGH SPEED CASCADABLE SIGNAL PROCESSING CIRCUITS FOR CASCADABLE MODULES TO PROCESS RANGE AND RANGE RATE DATA, REAL-TIME KALMAN FILTERING, REAL-TIME TARGET MOTION RESOLUTION (TMR) PROCESSING OF MRS-36 AND FPS-16 RADARS DATA, AND PROCESSING IMAGE/PATTERN INFORMATION FOR REAL-TIME TRACKING CONTROL OF OPTICAL TRACKERS. ALL THESE IMPROVEMENTS SHOULD ALLOW US TO INTEGRATE SEVERAL ARRAY PROCESSORS.