Simple eigenvalue-equalizing preprocessor for broadband power inversion array

By expressing the input autocorrelation function in a power series of bandwidth, this paper derives a small bandwidth first-order approximation for the covariance matrix of the broadband tapped delay line power inversion array. The structure of the associated eigenvalues is then deduced, giving (number of taps)/two sets of eigenvalues, the eigenvalues in the ith set being proportional to /(bandwidth) to the 2(i-1) power/. Thus, a simple transformation preprocessor which depends only on the bandwidth, tap spacing, and number of taps is derived for eliminating the spread of eigenvalues due to the use of tapped delay line processing.