SGD Frequency-Domain Space-Frequency Semiblind Multiuser Receiver with an Adaptive Optimal Mixing Parameter

Abstract

A novel stochastic gradient descent frequency-domain (FD) space-frequency (SF) semiblind multiuser receiver with an adaptive optimal mixing parameter is proposed to improve performance of FD semiblind multiuser receivers with a fixed mixing parameters and reduces computational complexity of suboptimal FD semiblind multiuser receivers in SFBC downlink MIMO MC-CDMA systems where various numbers of users exist. The receiver exploits an adaptive mixing parameter to mix information ratio between the training-based mode and the blind-based mode. Analytical results prove that the optimal mixing parameter value relies on power and number of active loaded users existing in the system. Computer simulation results show that when the mixing parameter is adapted closely to the optimal mixing parameter value, the performance of the receiver outperforms existing FD SF adaptive step-size (AS) LMS semiblind based with a fixed mixing parameter and conventional FD SF AS-LMS training-based multiuser receivers in the MSE, SER and signal to interference plus noise ratio in both static and dynamic environments.