Blind Signal Processing The Citic Source Number Estimation Method Study

In modern complex electronic communication environment, how to processing the mixed-signal of communication and radar rapidly and accurately has become one of the most difficult problems in communication reconnaissance.Blind source separation has been widely used in various fields recently because of its unique advantage and effective performance in the signal recognition and detection under the complicated signal environment. Source number needs to be estimated correctly for the successful realization of blind signal separation and extaction. The signal number estimation has been most used in array signal processing at present. It is more difficult to estimate in blind signal preccessing system as the lack of array manifold properties.Because the signal number determination is the crucial step in blind source separation, and the estimation performance will influence the whole system, the research on this domain is quite urgent and significant at present.The main work of this thesis is to research the communication and radar mix signal number determination in blind signal processing background. The key work and innovations of this thesis mainly include:1. Summarized the development process and related research status of blind signal processing and source number estimation. Several classic criteria are introduced in detail, then analyzed and compared with others in blind signal circumstance.2. Applying the multi-stage wiener filter techniqe under the wideband signal model into the blind source separation field. Simulations verify the good performance of the algorithm.3. Researched on the signal estimation algorithms base on eigen-decomposition. Considering the noise influence to the covariance matrix, the eigenvalue modification technique is studied to improve the performance in low SNR under over-determined situation.4. Reference on the array extension characteristics of high-order cumulant, the thorough study has been made on the channel extension technique with fourth order cumulant. Computer experiments of the underdetermined blind source estimation have been finished. The correct probabilities for detection have a better performance by utilizing the eigenvalue modification technique.Finally, an overview of author's work and the prospect of next research are presented.