Research On Equalization Algorithm Based On Factor Graphs For Underwater Acoustic Communication Systems

Underwater acoustic communication is highly valued in many countries because it is widely employed in national defense, civil marine development and etc.. Underwater acoustic communication technology is one of the most important part of the maritime project in our country. Extensive multipath, high attenuation, strong noise and limited bandwidth are characteristics of underwater acoustic (UWA) channels. Serious inter symbol interference (ISI) and low SNR at the receiver side result in the decrease of the reliability of underwater acoustic communication systems. It is a big challenge to communicate in high data rate and long range under this circumstance. The equalization algorithm based on maximum a posteriori (MAP) estimation is the optimal algorithm which is based on the minimum probability of the error criterion, while the high complexity of the algorithm restricts its practical application. Hence, approximation of the MAP algorithm and lower complexity technology had been developed. The factor graph emerged in90s which was widely used in signal processing provides a possibility to achieve low complexity and high performance equalization.This dissertation mainly studies the basic principle of the equalization algorithm based on factor graph. Changing the equalization to marginal problem by the sum-product algorithm, we detail the message passing method in the algorithm, and verify its performance by computer simulations. In order to meet the algorithm requirements for priori information of SNR, a new SNR estimation algorithm of a liner frequency modulation (LFM) signal based on fractional Fourier transmission (FRFT) is proposed. This algorithm is not sensitive to the change of the modulation scheme in communication systems and has high applicability. An iterative equalization structure combining LDPC with factor graph (LDPC-FG) is proposed based on basic factor graph equalization in order to increase the performance of the communication system at lower SNR. The results of Xiamen University experimental pool and the shallow water near Xiamen Port show that the proposed algorithm can ensure the reliability of the shallow water acoustic communication system.The main works of this dissertation are as follows:1. A new SNR estimation algorithm based on FRFT is proposed based on the different characteristics of the LFM signal and Gaussian white noise signal in FRFT domain.2. Factor graph and sum-product are introduced. How to use the sum-product algorithm to solve marginal functions in the Forney factor graph (FFG) and Gaussian message passing in this factor graph are introduced as well.3. By using the sum-product algorithm to fit the equalization into marginal problems and Gaussian message passing in factor graph framework to achieve the liner minimum mean squared error (LMMSE) estimation, the equalization based on factor graph is proposed and the performance of it is close to MAP.4. An iterative equalization structure combining LDPC with factor graph (LDPC-FG) is proposed and two implementations are presented in order to improve the performance of the communication system at low SNR.5. Results of experimental pool of Xiamen University and the shallow waters near Xiamen Port show that the proposed iterative algorithm combining LDPC with factor graph can effectively combat intersymbol interference in underwater acoustic channels, which could ensure the reliability of underwater acoustic communication systems.