CDMA system is a promising multiple access technique which extremely increases the channel capacity and supports multiple-access and multimedia transmission. However, the performance of CDMA system degrades in presence of multipath and dense multiple-access environment .Usually an adaptive filter is utilized to compensate the effect of multipath and multiple- access interference and signal fading over the channel. This paper provides a variable step size algorithm to accelerate convergence of the adaptive filter. The optimum performance is achieved by governing the rate of convergence and the amount of steady state excess mean square error (MSE).The performance is optimized in this
paper by two main methods mainly: variable step size parameters and gradient adaptively regularization parameter. The first method is implemented through variable step size least mean square algorithm VSSLMS, modified variable step size MVSS algorithm and variable step size affine projection algorithm VS-APA. On the other hand, the second method is
implemented using generalized normalized gradient descent algorithm GNGD and generalized square-error-regularized LMS algorithm GSER-LMS. This paper concerns with comparative study of the five mentioned adaptive algorithms,
(VSSLMS – MVSS – VS-APA – GNGD – GSER-LMS), to overcome the multipath and multiple-access problems of CDMA systems. The mentioned algorithms had the advantages of fast convergence and low steady state MSE, these advantages candidate the mention algorithms to improve the bit error rate (BER) performance of the CDMA system, in the presence of multipath and multiple-access environment. The results indicate that the proposed algorithm provides faster convergence compared with the fixed step size and fixed regularization parameters algorithms.