An Underwater Acoustic Channel Equalization Algorithm Based on RBF Neural Network

With the increasing demand for the performance of communication transmission system, new requirements are put forward for channel equalization technology. There are serious multipath interference and Doppler phenomena in the underwater acoustic channel. In view of this phenomenon, equalization technology is proposed to suppress it. The application of machine learning in image processing, machine vision, communication and other fields has achieved remarkable results. Based on this, this paper proposes a channel equalization algorithm which combines intelligent optimization algorithm and machine learning network. This algorithm utilizes the PSO (Particle Swarm Optimization) algorithm to optimize the RBF (Radial Basis Function) neural network and adjust its parameters for signal equalization under the underwater acoustic channel. The algorithm takes a QPSK (Quadrature Phase-Shift Keying) signal transmitted through the underwater acoustic channel as input to the network. During the adjustment of network parameters, the PSO algorithm is employed with a population size equal to the number of parameters requiring optimization. Simulation results have demonstrated that using this algorithm for equalization outperforms using a simple neural network. The constellation is more concentrated, and the error rate is reduced by approximately 5 %.