A Data-aided Compensation of Clipped OFDM Signals by Bayesian Learning Scheme

Abstract

Clipping is a simple technique for the peak-to-average power ratio reduction of orthogonal frequency division multiplexing (OFDM) systems, and the clipping signal recovery is required at the receiver. The nonlinear distortion is a sparse phenomenon in the time domain, and a compressed sensing technique is proposed to compensate for the amplifier's distortion for a coded OFDM system. This scheme involves an iterative data-aided algorithm that does not require any pilot carriers, and the clipping signal recovery depends on a subset of reliably detected subcarriers that the channel coding is taken into consideration. In particular, a sparse Bayesian learning method is performed for clipping signal recovery, where the parameters of the signal distribution can be learned through the expectation maximization. Numerical experiments demonstrate the favorable performance for the proposed technique compared to other sparse reconstruction methods.