Sparse channel estimation and tracking for cyclic delay diversity orthogonal frequency division multiplexing systems

In cyclic delay diversity orthogonal frequency division multiplexing systems, the excessive channel delay spread and corresponding high frequency selectivity makes channel estimation a challenging task. In this paper, we propose a two-stage scheme to estimate and track the highly frequency selective channel. At the preamble reception stage, least squares channel estimation with L0 norm regularization is proposed to exploit the channel sparsity. At the data demodulation stage, an expectation–maximization algorithm with the most significant tap selection is developed to track channel variations by using the channel order obtained from the first stage. Compared with other estimation methods, the proposed scheme requires no prerequisite knowledge of delay parameter settings, which leads to more flexibility. Furthermore, the scheme can exploit the channel sparse structure by detecting the nonzero taps and, consequently, has better mean squared error performance. Simulation results show that the proposed estimation scheme can retain the provided diversity gain of cyclic delay diversity effectively in time-varying fading channels. Copyright © 2011 John Wiley & Sons, Ltd.