Abstract:SNR estimation of communication signals is important to improve demodulation performance and channel quality of communication system, thus it is an important research issue of communication field. According to the core problem of autocorrelation matrix singular value in SNR estimation process, through making use of householder transforming autocorrelation matrix into tridiagonal matrix, and by using the relation of corresponding characteristic equation coefficients and singular value, a numerical algorithm is given to obtain autocorrelation matrix singular value, and the algorithm is used for SNR solving process. Theoretical analysis shows that the algorithm can satisfy the requirements in the aspect of constringency speed and stability. This work has been supported by the National Natural Science Foundation of China (Grant No. 90604031)