STO Estimation for OFDM System Using CDM

Abstract

In this letter, we propose a novel scheme for symbol timing offset (STO) estimation by using a convolutional neural network (CNN)-deep neural network (DNN) model (CDM) architecture for the orthogonal frequency division multiplexing (OFDM) system over different fading channel models. The proposed scheme estimates STO in the presence of carrier frequency offset (CFO) and without prior knowledge of the channel, modulation format, and transmitted OFDM signal parameters. The CDM architecture achieves better estimation accuracy gain compared to that of statistical-based and pilot-assisted CNN-based methods. Finally, the proposed CDM is validated over a radio frequency testbed, and the desired constellation diagram is obtained.