Comparing the radar clutter-suppression performances of lattice prediction error filters using three variations of Burg´s algorithm

An adaptive technique is presented for suppressing clutter in a moving target indicator radar signal processor. The technique is based on the assumption that radar clutter can be modeled as a Mth-order auto-regressive (AR) process and employs a lattice prediction error filter structure to suppress clutter without significantly distorting the target signal. Filter coefficients are the reflection coefficients determined using one of the three variations of Burg´s AR parameter estimation algorithm due to Andersen, Makhoul and Marple. These filter designs are applied to second-order Butterworth-shaped spectra representing both ground and storm clutter, and their performances are compared with each other and with more conventional methods in terms of improvement in signal-to-clutter ratio between filter input and output. Results are presented which show the dependence upon such factors as sample size, filter order, and doppler frequency.