The blind equalization and multiuser detection problem for synchronous high-rate DS-CDMA systems in dispersive channels was considered. Using only the spreading code of the desired user and the property that the transmitted signals belong to the finite discrete-alphabet

the problem was formulated into a quadratic optimization with binary constraints. Then

an efficient ε-approximation algorithm was applied to find the solution of the quadratic optimization without channels estimation and designing the coefficients of the equalizer explicitly. This approach could reduce the performance degradation that was caused by channels estimation errors and have the superiority over the varying channels. Simulations demonstrate its superior performance to the subspace method and constant modulus algorithms.