The approach for design of parallel implementation of the second-order Kalman filter using systolic array processors is proposed. The original time complexity of more than 27n/sup 3 /for sequential processing can be reduced to only 21n+10r-4 in the proposed parallel implementation. The improvement is gained from the utilization of interarray pipelining, concurrent processing and substitution forwarding techniques. The approach is especially suitable for real-time application of the Kalman filter in solving second-order linear system.
展开▼