Fixed window constant modulus algorithms
Xiang-Yang Zhaung, Arnold Lee Swindlehurst · 1999
We propose two batch versions of the constant modulus algorithm in which a fixed block of samples is iteratively reused. The convergence rate of the algorithms is shown to be very fast. The delay to which the algorithms converge can be determined if the peak position of the initialized global channel/equalizer response is known. These fixed window CM algorithms are data efficient, computationally inexpensive and no step-size tuning is required. The effect of noise, and the relationship between the converging delay and noise enhancement are analyzed as well.