DNN Based Iterative Detection for High Order QAM OFDM Systems With Insufficient Cyclic Prefix

In this paper, we consider high order QAM orthogonal frequency-division multiplexing (OFDM) systems with insufficient cyclic prefix (CP) which will lead to intersymbol interference (ISI) and intercarrier interference (ICI) in the receiver. To cope with the error floor induced by ICI and ISI in one-tap equalization and iterative serial interference cancellation (ISIC), we develop a maximum likelihood (ML) based iterative grouping detection algorithm (ML-IG), which utilizes the correlation among the received signals on adjacent subcarriers to improve the detection accuracy. Since ML-IG for OFDM systems with high order QAM modulation is of significant complexity, detection network (DetNet) based iterative grouping detection algorithm (DetNet-IG) is designed to imitate ML-IG. Simulations show that both ML-IG and DetNet-IG can provide better BER performance than ISIC, and DetNet-IG exhibits distinguished robustness against channel model incongruity.