Deep neural network-based intercarrier interference detection for optical spectral efficient frequency division multiplexing system

Abstract

In the past few decades, many efforts have been made to solve the intercarrier interference (ICI) because of the loss of the orthogonality for the spectral efficient frequency division multiplexing (SEFDM) optical communication systems. In this paper, a deep neural network (DNN) is introduced to deal with the ICI. The intrinsic relationship between the mechanism of ICI damage and the DNN is studied and analyzed. Based on this analysis, the performance of DNN-ICI decoder compared with the conventional algorithms is demonstrated by simulation for optical SEFDM intensity modulation/direct detection (IM/DD) communication systems. The results show that the DNN-ICI decoder is greatly superior to other schemes in terms of bit error rate (BER) with a simple designed DNN under different bandwidth compression factors. Besides, the proposed methods are also robust to the fiber lengths. All the results indicate that the proposed DNN-ICI decoder has great potential to be used in SEFDM IM/DD optical systems.