Neural Network Approach for Underwater Acoustic Communication in the Shallow Water

The transmitted acoustic signals are severely influenced by boundaries like as sea surface and bottom in the shallow water. Very large reflection signals from boundaries cause inter-symbol interference effect, the performance of the communication are degraded. Usually, to compensate the reflected signals under this kind of acoustic channel, is adopting the channel estimation based equalizers. In this study, we express neural network approaches for image data transmission in the shallow water. A simple neural network is adopted for the decision from output data. The QPSK system is used for the underwater acoustic communication simulations and experiments.