Orthogonal chirp division multiplexing (OCDM) offers a promising modulation technology for shallow water underwater acoustic (UWA) communication systems due to multipath fading resistance and Doppler resistance. To handle the various channel distortions and interferences, obtaining accurate channel state information is vital for robust and efficient shallow water UWA communication. In recent years, deep learning has attracted widespread attention in the communication field, providing a new way to improve the performance of physical layer communication systems. In this paper, the pilot-based channel estimation is transformed into a matrix completion problem, which is mathematically equivalent to the image super-resolution problem arising in the field of image processing. Simulation results show that the deep learning-based method can improve the channel distortion, outperforming the equalization performed by traditional estimator, the performance of Bit Error Rate is improved by 2.5 dB compared to the MMSE method in OCDM system. At the 7.5 to 20 dB region, it achieves better bit error rate performance than OFDM systems, and the bit error rate is reduced by approximately 53% compared to OFDM when the SNR value is 20, which is very useful in shallow water UWA channels with multipath extension and severe time-varying characteristics.

In this paper, a partial transmit sequence (PTS) technique with a side information (SI) blind detector based on minimized error accumulation for PAPR reduction in the coded underwater acoustic (UWA) orthogonal frequency division multiplexing (OFDM) communication system is proposed. Due to the complexity of UWA channels, channel coding is inevitably employed to ensure accurate information transmission. However, the error correction capability of channel coding usually has a certain margin; the decoder will fail to provide the expected coding gain for the system when the number of error bits exceeds the error correction capability. Therefore, by utilizing the accumulation impact of bit errors inherent in channel coding, it can be regarded as an indicator vector for the SI blind detector, enabling autonomous identification of the weighted phase factor vector sequence index without prior information. Thus, there is no need to reserve the SI transmission symbols, as required by the conventional PTS (C-PTS) scheme, effectively reducing the probability of large-scale bit error caused by the wrong SI. The experimental results show that the BER performance of the system has seen an improvement of about one order of magnitude compared to C-PTS. Consequently, the proposed technique enhances system bandwidth utilization and communication efficiency, ensuring the real-time performance of the underwater acoustic communication (UAC) OFDM system.