Blind adaptive detection using differential CMA for CDMA systems

Abstract

In this paper, we consider the application of adaptive linear filters adjusted by stochastic gradient algorithms to the suppression of multiple‐access interference in DS/CDMA communications. It is pointed out that the linearly constrained constant modulus algorithm (LCCMA) cannot converge to the optimum point if the desired user magnitude is less than a critical value. Next, we propose to use the linearly constrained differential constant modulus algorithm (LCDCMA), and show that the LCDCMA always converges to the optimum point regardless of the desired user magnitude. Several simulation results show the following three points. First, the LCDCMA provides better performance than existing algorithms because the variance of the weight vector adjusted by the LCDCMA is less than that by existing algorithms. Second, the LCDCMA is relatively insensitive to the estimation error of the desired signal vector. Third, the LCDCMA combined with a blind channel estimation algorithm is useful. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 85(1): 1–13, 2002

Steady-state performance analysis of HOS-based blind adaptive multiuser detection

“…The DCM multiuser detection has also the advantage of no existing local minima [5] . It has been proved by simulation that the DCM algorithm may provide good decorrelation performance in high SNR environment, whereas a theoretical analysis is lacking.…”

“…Not only is it able to effectively suppress the colored noise, but also can obtain higher performance of decorrelation than the second-order statistics (SOS) methods in high signal-to-noise ratio (SNR) environment [2] . To ensure the convergence and decrease the computing complexity, several new HOS-based multiuser detection algorithms have been proposed, including the least mean kurtosis algorithm (LMK) [3,4] , the differential constant modulus (DCM) algorithm [5] , and the least mean fourth algorithm (LMF) [6] .…”

Performance analysis of the linearly constrained constant modulus algorithm-based multiuser detector

“…The orthogonal projection of the received vector (6) is used in (5) to ensure that remains orthogonal to , which in turn ensures that the constraint is always met. The noncanonically constrained LCCMA was proposed in [16] and further analyzed in [4]; the variation sets , and the filter coefficients are decomposed into (7) subject to the linear constraint , and has the stability constraint . It is important to note that the cost functions of the noncanonically constrained LCCMA and canonically constrained LCCMA are equivalent when , [17].…”

“…The constant modulus algorithm (CMA) [2] has recently been applied to the blind adaptive MUD problem, where the various versions [linearly constrained (LC-) CMA [3], [4], multiuser (MU-) CMA [5], cross correlation (CC-) CMA [6], and linearly constrained differential (LCD-) CMA [7]] have modified the CMA cost function to ensure that the desired user is captured. The LCCMA scheme is considered in this paper because of its relative simplicity and proven resilience under mismatch conditions [3].…”

The Linearly Constrained LSCMA for Blind Multi-user Detection

“…But the conventional CMA can easily capture the interfering signals other than the expected one in the CDMA system where there exist many constant modulus signals. For improvement on the tracking capability, the linearly constrained CMA [11][12][13] and the differential CMA [14] are reported, but their convergence are more easily influenced by the step-size.…”