Research On Source Number Estimation Methods

Estimation of the number of signal sources is an important branch in array signal processing, and has important theory value in the medicine, the correspondence, the radar, the sound navigation and ranging, the seismic survey, the pronunciation, the image and the finance and economics and so on. Estimation of the number of signal sources has become a main development direction of signal processing domain. In these domains, efficienct estimation of the number of signal sources is very important. Especially in the signal processing domain, it has a very vital role to the signal processing. The existing algorithms request the number of signal sources to be equal or to be smaller than the number of array elements generally, and hope that the number of signal sources is known, but in fact the latter is very difficult to achieve. In the treating processes, people suppose the number of sensors and the number of signal sources generally is equal. Therefore, estimation of the number of signal sources has an important meaning to technological development in the array signal processing, and is also the question which must give to solve at present.This paper first describes estimation methods of the number of signal sources in the condition of the white noise, and then focuses on estimation methods of the number of signal sources in the condition of the color noise. The main work is listed as follows:(1) In the condition of the white noise, estimation methods of the number of signal sources based on information theory criteria are studied. Paper mainly discusses the context of AIC, MDL, EIT, RAIC in the white noise such as the principle and algorithm implementation steps. AIC and MDL criteria are studied through simulation and analysis in order to find the relationships between the two methods based on the same angle, SNR, sample size, the number of array elements and so on.(2) The classic algorithm-Gail circle criteria is introduced, this paper proposes Gaelic circle criteria based on the pseudo-covariance matrix in the colored noise. Simulation experiments reveals that the Method performance is better than other similar algorithms.(3) For the question that the colored noise make the array covariance matrix characteristic values divide, this paper carrys out in-depth analysis and studys the diagonal loading techniques, then focuses on the question of loading volume.(4) For the existing algorithm has bad performance problems in colored noise environments, the paper presents a kind of source estimation method based on eigenvector. The method in white noise and colored noise cases are both valid, and in particular in the case of colored noise. Through the simulation it is found that this method is superior to other similar algorithms in the low SNR and small snapshots. The method is consistent estimates, robust, strong, and the computation is not great.