Fuzzy-logic tuned constant modulus algorithm and soft decision-directed scheme for blind equalisation

Abstract

This contribution revisits blind equalisation for high-order quadrature amplitude modulation systems using a low-complexity high-performance concurrent constant modulus algorithm (CMA) and soft decision-directed (SDD) scheme. A fuzzy-logic (FL) tuning unit is designed to adjust the step size of the CMA, and the potential benefits of using this adaptive step size approach are investigated. Simulation results obtained confirm that faster convergence can be achieved with this FL assisted CMA and SDD scheme, compared with the previous constant step-size CMA and SDD scheme. Quantitatively, the former requires several thousands fewer samples to converge to the same steady-state solution achieved by the latter.

Fuzzy step size least mean square algorithm tuned linearly constrained constant modulus algorithm for multi‐carrier code division multiple access blind multiuser detector

“…Hence, as mention before, a variable step size is necessary to acquire not only the rapid convergence speed but also the low steady state fluctuation error. Therefore, an FL step size LMS algorithm is proposed in , which employs an FIS system with the use of two inputs e 2 ( i ) and i to adjust the step size, denoted as the positive scalar μ ( i ), where e 2 ( i ) denotes the square of e k and i denotes the duration of training, the algorithm is named as FL step size training duration LMS algorithm (FSST‐LMS): …”

“…Therefore, the next step in the FIS is to find the proper step size by aggregating all the outputs from the membership function and the step size adjustment table. The centroid calculation is employed here : where q is the output numbers of the membership function, μ j ( i ) is the step size obtained from Tables and , and m B ( μ j ( i )) ∈ [0,1] is the membership function value of μ j ( i ).…”

“…For a more consistent expression, we can denote the FSST-LMS as FLC-LMS (e 2 , i). The fuzzy step size LMS filter developed in [23][24][25][26][27] employs an FIS with the use of two inputs e 2 (i) and i. However, for the time varying wireless communication, the FL step size LMS filter, which uses e 2 (i) and i, cannot efficiently track the channel variation.…”

Invasive weed optimization method based blind multiuser detection for MC-CDMA interference suppression over multipath fading channel

“…[11], Chern et al [12] , Kimet al [13] and Sanchez et al [14], and S. Chen et al [20] study that CMA blind multiuser detection is applied in MC CDMA systems. Van Meng et al [11] and J. Namgoong [15] suggests that subspace-based linear MMSE detection be used in MC DS-CDMA systems.…”

“…A constant step-size LMS algorithm thus has to trade off between the steady-state performance and convergence speed when choosing the step size value. In attempts to optimize both the steady-state performance and convergence speed, techniques based on fuzzy logic (FL) tuning of LMS' s step size have been developed [16][17][18][19][20][21]. The CMA is a stochastic gradient blind adaptive algorithm, and its step size has to be chosen with extreme care, much more than the training-based LMS algorithm.…”