Abstract: ‍ ‍In this paper， the problem of blind de-expansion of short code direct spread signal in non-cooperative spread spectrum communication is studied. Aiming at the problem that the out-of-step time estimation of existing methods is not accurate under the condition of low SNR， an improved blind estimation method of direct spread signal information and pseudo-code sequence based on eigenvalue decomposition is proposed in this paper. Methods According to the characteristic of the eigenvalue decomposition of the first kind of received signal autocorrelation matrix， the out-of-step point estimation is completed by using the relationship between the eigenvalue of the matrix and out-of-step point， and then the information sequence is estimated by using the eigenvector when the pseudo-code sequence is unknown. After the out-of-step point estimation is completed， the blind estimation of the pseudo-code sequence is realized by using the eigenvectors of the second type of received signal autocorrelation matrix. Experimental results show that the method has the advantages of high estimation accuracy， good stability and strong anti-noise performance.