Novel variable step size constant modulus algorithms for blind multiuser detection

Conventional constant modulus algorithms (CMAs) suffer shortcomings of slow convergence rate and possibility of improper convergence, which obstruct their application in real-time multiuser detection for CDMA cellular system. This paper focuses on developing new algorithms to alleviate these problems. Three new CMAs, recursive variable step size CMA, normalized constrained CMA and recursive variable step size constrained CMA, are obtained under a uniform framework. Interestingly they can be expressed in a general form. Since they automatically and adaptively choose the optimal variable step size there is no need to worry about whether the step size works and they converge much faster. To solve the modulus parameter sensitivity problem of constrained CMA, modulus parameter correction is integrated into RVSS-CCMA. Various simulations validate the efficiency and robustness of our new algorithms, especially the RVSS-CCMA with modulus parameter correction