The design of multiple orthogonal signals based on auto-encoder optimization network

As an important part of active sonar, transmitted signals have a great influence on the performance of ocean exploration, however, in the actual environment, due to the existence of the Doppler frequency shift, the traditional transmitted signals may have a relatively bad performance. Therefore, in a bid to improve the adaptability of the transmitted signals in the actual environment, a signal design method has been proposed in this paper. In this method, a neural network based on auto-encoder has been presented and a randomly normalized phase sequence is sent into the network and then it is optimized by minimizing the ambiguity function peak side-lobe level of the phase sequence. Compared with the existing methods, the network structure is adjusted to further consider the transmitted signal optimization under the Doppler frequency shift. Simulation results show that the optimized signals perform superior than the existing method under Doppler frequency shift, which may improve the performance of ocean exploration to some extent.