The design and analysis of adjustment factor in Gerschgorin Criterion for Source Number Estimation

This paper analyzes and studies the influence of different adjustment factor upon the algorithm property in Gerschgorin Criterion for Source Number Estimation. Then a new design method of adjustment factor is proposed, which is irrelevant to sampling points. This adjustable factor-designed don't need be chose subjectively, and it is determined completely by centre information of Gerschgorin circle. The performance of algorithm using this adjustable factors are discussed and compared through computer simulation including GDE algorithm and several improve GDE algorithm in the different SNR and sampling points under colored noise. Results show: overall, performance of GDE algorithm using adjustment factor designed by this paper is better than that used other adjustment factor, and performance of source number is very good when sample size is little.