Optimum finite memory nonlinear filter
Ching-Chyoun Yang · 1965
This thesis investigates the possibility of using the Zadah theory and a machine type computer to desigp an optimum :finite memory nonlinear filter :for extraction o:f ]?oisson distributed signal pulses :from a backg~ound of' Guassian noise. Basically the procedure involves the technique of solving simultaneous equations by the computer. We demonstrate that it is possible to find the limitation o:f the :filter o:f classrl1 in terms o:f an absolute minimum mean square error. However, due to the capacity of the computer we do not actually determine the absolute minimum. Nevertheless we do show that the mean square error decreases with increasing memory time T.