New Combined Adaptive (NCA) Algorithm for Smart Antenna System Applications

Abstract

ABSTRACTThis paper proposes a new combined algorithm for performance enhancement of smart antenna systems in mobile communications. The new combined algorithm (NCA) uses sample matrix inversion (SMI) initialization with a modified version of normalized least mean square (NLMS) algorithm (MNLMS). The MNLMS algorithm is regarded as time-varying regularization parameter ϵ(k) that is fixed in the conventional NLMS algorithm. The MNLMS algorithm uses a reciprocal of the estimation error square term in the denominator of the learning rate of NLMS instead of fixed regularization parameter (ϵ). Through simulation results of smart antennas for an additive white Gaussian noise channel model, the NCA algorithm achieves about 15 and 19 dB improvement in interference suppression compared with NLMS and another proposed previously matrix inversion NLMS (MI-NLMS) algorithms, respectively. While the NLMS algorithm starts to converge from 50 iterations, the MI-NLMS and NCA algorithms start to converge from the initial iteration. Moreover, when this algorithm applied for the Rayleigh fading channel with a Jakes power spectral density, it provides about 7 dB improvements over NLMS and MI-NLMS algorithms and starts converge from the initial iteration. The NCA provides improved performance in weight stability for both channels.