Communication signals are appropriately to be modeled as cyclostationary stochastic processes. The most significant advantage of cyclostationarity is its suppression on the stationary noise and interference. Therefore, cyclostationary theory has been a powerful tool in LPI (Low Probability of interception) signal processing. In non-cooperative condition, blind detection and parameter estimation of DS-CDMA (Direct Spread Spectrum code division multiple access) signals is the first step of PN (Pseudo Noise) codes extraction and information recovery, which has been a challenging problem for a long time and has much theoretical and practical values. The main work of this paper is to deal with multiuser DS-CDMA signals in multipath entironment. Method based on second order cyclostationary statistics was adopted to detect whether the modulated signal exists in background noise. The close formula of SCF (spectrum correlation function) for complex DS-CDMA signals was derived, chip rate can be achieved from the section of SCF where alpha = f0 (alpha is the cyclic frequency and f0 is the carrier frequency ). Simulation results containing the alpha = f0 section and the ROC (Receiver Operating Characteristics) curve have been shown, which proves that the proposed method can achieve good performance even when signals are buried in strong background noise.