A Method of Multipath Mitigation Based on LS and Complex ResNet in Water Supply Pipes

There are many challenges in underwater acoustic communication systems, as the underwater acoustic channel characteristics are highly dependent on the water environment. The multipath effect is one of the most impactful factors to the performance of communication. Water supply pipes are common structures in underwater facilities and widely used on the Internet of Underwater Things (IoUTs). Underwater acoustic (UWA) communication is occasionally needed in IoUTs applications using pipes, like submarine pipeline monitoring, but there is little research about communication in water supply pipes. This paper optimizes the underwater acoustic channel estimation for the water supply pipes channel. A complex residual network (ResNet) based on least square (LS) is introduced to mitigate the multipath effect in water supply pipes. The proposed method shows better performance in channel estimation than the conventional LS method and performs almost as well as the minimum mean square error (MMSE) method. The network is tested in water supply pipes by using an OFDM communication system. The simulation result substantiates the validity of the proposed optimized method in a pipe communication system.