Abstract
Adaptive beamformers balance the competing demands of suppressing strong undesired signals by creating destructive interference while attenuating uncorrelated background noise. The hybrid minimum variance distortionless response (HMVDR) beamformer for narrowband passive sonar achieves this balance by factoring its beampattern as a product of a conventional delay-and-sum beampattern and an adaptive Capon beampattern. The conventional beampattern attenuates uncorrelated background noise, while the adaptive beampattern steers nulls to cancel loud interferers. Multiplying the beampatterns corresponds to convolving two smaller weight vectors for notional subapertures. The two factors are chosen so the total number of degrees of freedom (DoFs) in the product equals the number of sensors in the array. Universal methods allow the hybrid beamformer to adapt the partition of the DoFs between the two factors to respond to changes in the environment. Simulations and microphone array data demonstrate that the HMVDR beamformer performs better than the standard Capon adaptive beamformer or the covariance matrix taper. [Work funded by ONR 321US.]