Fading channel prediction based on complex-valued neural networks in frequency domain

Channel prediction is an important process for channel compensation in fading environment. If a future channel state is predicted, adaptive techniques such as pre-equalization and transmission power control is applicable before transmission in order to avoid degradation of communication quality. Previously, we proposed frequency-domain channel prediction methods employing the chirp z-transform with a linear extrapolation as well as a Lagrange-based nonlinear extrapolation of frequency-domain parameters. This paper presents a highly accurate method for predicting time-varying channels by using a complex-valued neural-network (CVNN) based prediction of frequency-domain channel characteristics. We demonstrate that the channel prediction accuracy of our CVNN method is better than those of the previous linear- and Lagrange-extrapolation prediction in a series of simulations for orthogonal frequency division multiplexing (OFDM) communications.