Adaptive radar detection in non-stationary Doppler spread clutter

The detection performance of a radar depends on its ability to distinguish a weak target from strong surface backscatter clutter in the Doppler frequency space. Under stable propagation conditions, conventional Doppler processing can discriminate moving targets from nominally stationary clutter. However, in complex multipath propagation scenarios where path lengths can change abruptly during the radar coherent processing interval (CPI), the backscattered clutter return is spread in Doppler space thereby masking the presence of weak targets. Furthermore, non-stationarity of clutter across range bins (i.e. fast time) severely limits the amount of training data available to estimate the adaptive processor weights. This thesis addresses the problem of radar detection in clutter which is non-stationary in both slow and fast time. Our approach models the clutter return as a modified time-varying autoregressive (TVAR) process that accounts for both slow and rapid variations in the clutter statistics. To overcome the lack of training data, TVAR model based maximum likelihood estimate of the clutter covariance is computed from single snapshot of radar data. The proposed adaptive detector is an approximation to the generalized likelihood ratio test (GLRT) wherein the CPI from the hypothesized target range is used for both clutter estimation and target detection. Experimental simulations using over-the-horizon (OTH) radar data with injected targets show that the GLRT detector provides around 6 dB improvement in the output signal to clutter plus noise ratio (SCNR) compared to the conventional processor.; In this thesis, we also present a physics based ionospheric propagation model for pursuing joint spatio-temporal clutter mitigation strategies. Particularly, in equatorial regions, motion of ionospheric irregularities can cause significant Doppler and azimuth spread in the received clutter. The proposed approach assumes that much of the Doppler spread clutter arrives via higher elevation raypaths while less Doppler spread targets are detected on lower elevation raypaths. This thesis presents a simplified propagation model that employs thin phase screen techniques for modeling the first order features of radar clutter. Clutter observations recorded by an equatorial OTH radar show that our model can predict very well the dominant features in the ionospheric scattered clutter returns.