Application of FastICA Algorithm in the Fast-scanning Radar
Meng Qiu-chi, Ziliang Zhang · Modern Radar · 2012
To Modern Doppler radar,calculation accuracy of the data depends on the number of independent samples,the more independent samples the more accurate in the calculation.In order to enhance the accuracy and improve the speed of radar scanning, in this paper,independent component analysis technique is used to increase the number of independent samples.ICA algorithm is on the basis of kurtosis,maximum likelihood estimation or other mathematical methods.It estimates separating matrix gradually through the iterative approximation so that the independent factors can be achieved.Among those algorithm,fast fixed-point algorithm can improve the speed of calculating by online learning and enhance its reliability.In this paper,a fast fixed point algorithm based on the maximum likelihood estimation algorithm is applied to rayleigh signals and simulative radar signals respectively and the results are compared with principal component analysis which is commonly used.The results of the texts prove a good performance of ICA which fully verifys that ICA techniques could more accurately estimate independent signal from the mixed-signal, greatly increasing the number of independent samples to reduce the spectral estimation errors.This has laid the foundation for the practical application of Doppler radar signal processing.