Parameter estimation based on factor graph in wide‐band OFDM systems

Abstract

The channel estimation in a wide-band orthogonal frequency division multiplexing (OFDM) system is different from the narrow-band system, for the estimation parameters should include the Doppler scale factor, which is a particular parameter of the wide-band system. The underwater acoustic OFDM system is a typical wide-band system, and its channel has more than one scale factor usually. For this multi-scale case, the common estimation method is a factor-by-factor process with repeated calculation of the reproduced waveform. This article proposes a factor-graph-based estimation method, which fully uses the restriction relationship among the parameters, and estimates the parameters simultaneously. To reduce the computation load, the modified Bayesian propagation (BP) algorithm is proposed. Comparing to the common BP algorithm, the modified BP algorithm simplifies the calculation of the condition probability in the summation. The simulation results of the mean squared error show that the proposed method is an effective estimation method for the multi-scale wide-band system. The bit error rate performance of the proposed method is very close to the result of the known channel-state-information, and outperforms the fractional Fourier transform (FrFT) and the modified particle swarm optimisation (MPSO) methods for well channel condition.