Superimposed Training-Based Channel Estimation for MISO Optical-OFDM VLC

In this paper, we investigate a novel channel estimation (CE) method for multiple-input and single-output (MISO) systems in visible light communication (VLC). Direct current biased optical orthogonal frequency division multiplexing (DCO-OFDM) is commonly used in VLC where half of the available subcarriers are spent to guarantee a real-valued output after the inverse fast Fourier transform operation. Besides, dedicated subcarriers are typically used for CE, thus, many resources are wasted and the spectral efficiency is degraded. We propose a superimposed training approach for CE in MISO DCO-OFDM VLC scenarios. Analytical expressions of mean squared error (MSE) and spectral efficiency are derived when the least squares estimator is considered. This analysis is valid for outdoor and indoor scenarios. For the CE error, simulation results of MSE show a perfect match with analytical expressions. Moreover, results prove that this technique guarantees a larger spectral efficiency than previous schemes where dedicated pilots were used. Finally, the optimal data power allocation factor is also analytically derived.