Adaptive beamforming with parametric estimation of the correlation matrix

It is well known that correlation between interferers and the signal arriving from the look-direction severely degrades the output SINR for both low-rank and full-rank minimum variance distortionless response (MVDR) beamformers. Low-rank MVDR beamformers, such as conjugate gradients (CG) and principal components inverse (PCI), can yield a higher SINR than full-rank MVDR when the sample support is inadequate. However, a large drop in output SINR occurs if the low-rank beamformer operates at an improper rank. Indirect dominant mode rejection (IDMR) is proposed wherein one first employs a high-resolution spatial spectrum estimation technique to estimate the directions and powers of the dominant interferers. Subsequently, this information is used to construct an estimate of the signal-free (interference plus noise only) autocorrelation matrix (for a given look-direction.) In this process, any correlation between the interferers and the signal arriving from the look-direction, due to either finite sample averaging or true correlation amongst signals, is effectively removed. Simulations reveal that IDMR is very robust to correlation amongst signals and yields a dramatic improvement in output SINR relative to CG and PCI/DMR, even when the latter operate at the optimal rank.