Abstract: A Constrained least-mean-square (LMS) algorithm is proposed, which is used for an adaptive optimum combination(MMSE criterion) of diversity signals in orthogonal frequency-division multiplexing (OFDM) systems. Compared with the conventional algorithm, the proposed algorithm has no drawback of slow convergence speed when the input signals are highly correlated. Before the received diversity signals are fed to the adaptive filter, they are transformed to have the identical desired response. Then, the transformed signals are optimally combined in the adaptive filter subject to the proposed constraints. The derived analytical expressions indicate that regardless of the input signal correlation matrix, the convergence behavior of the proposed algorithm, in all cases, is only governed by the normalized step size.

Key words: adaptive optimum combining, LMS algorithm, OFDM