Toeplitz constrained adaptive beamforming with auxiliary data

Abstract

Submarine and mine detection using active sonar face similar problems when adaptive beamforming is required to provide high angular resolution or to suppress angularly isolated interferences. Both

situations result in a very small amount of target-plus-interference data and a moderate amount of

stationary interference-only data. From this data the target-plus-interference arid interference-only

spatial covariance matrices (SCM's) must be estimated. When the SCM's have a Toeplitz structure,

as will occur with an equi-spaced line array and plane wave propagation, beamforming performance

may be improved by constraining the estimates to be Toeplitz as well. In this report, the expectation maximization algorithm is used to obtain the Toeplitz constrained maximum likelihood estimates of the

SCM's when target-plus-interference data and interference-only (auxiliary) data are available. When

the target-plus-interference SCM estimate is used in an MVDR beamformer, it is shown that the beam

output SNR is non-linearly dependent on the maximum beam output SNR (Smax) in such a manner that when Smax is large, accurate estimation of the SCM is crucial. Simulation analysis illustrates that

the proposed algorithm can provide the necessary accuracy, with further improvement as the amount of auxiliary data increases.