A New MSE Channel Estimator Optimized for Nonlinearly Distorted Faded OFDM Signals With Applications to Radio Over Fiber

In this paper, we propose a new channel estimation (CE) structure that aims to minimize the mean square error (MMSE) of orthogonal frequency-division multiplexing signals that are corrupted by fading, nonlinear, and additive white Gaussian noise channels. We demonstrate that it exhibits a simple structure consisting of a set of correlators matched to every frequency component in the nonlinearly distorted signal, followed by a linear transformation. An extension suitable for frequency shifted signals is also discussed. As case study, the potential of such CE in uplink radio-over-fiber (RoF) systems conveying WiFi signals is analyzed. Through simulations, we demonstrate the usage of the one-tap and iterative maximum likelihood (ML)-algorithm-based equalizers, when used jointly with the proposed MMSE, zero forcing, or perfect CEs. Results show that the performance of the MMSE CE approximates the one of the perfect CE. In addition, when used in conjunction with the ML algorithm, it enables relaxation of linear requirements of components in RoF by passing the need for transmitter predistortion techniques that would otherwise significantly increase wireless devices complexity.