Estimation of source number under conditions of mutual coupling effects and channel amplitude mismatch

The MDL criterion and AIC criterion based on information theorem have excellent performance under the condition of uniform Gauss white noises. But when there are the mutual coupling effects and channel amplitude mismatch in the system, the noise changes to a non-uniform and colored noise. And under this condition, it is unpossible to estimate the source number by using the MDL criterion and AIC criterion yet. To deal with this problem, the inner-relationsheep between the information theoretic method and Gerschgorin radius method and chooses Gerschgorin radius as the parameter of the probability model is analyzed. And then two improved criteria named GMDL criterion and GAIC criterion are proposed. The computer simulation and measured data results illustrate the excellent performance of the proposed GMDL criterion and GAIC criterion.