Research On Channel-estimate-based Turbo Equalization Algorithms In Underwater Acoustic Communication

Ocean with abundant marine resources is a frontier to be developed compared with land.Underwater acoustic communication is an important tool for exploring and exploiting marine resources,which is facing challenges on several difficulties.Complex underwater environment inducing long-tap multipath channels,can directly influence the reliability of the communication performance.To counteract the multipath effect and inter-symbol interference,turbo equalization becomes a hot research now.This paper focuses on channel-estimate-based turbo equalization(CE-TEQ)algorithms applied in underwater acoustic communication.The main research contents in this paper are as follows:First,an statistical characterization and computationally efficient underwater channel model is researched.Large-scale and small-scale random channel variations are considered to build the channel model which incorporates statistical characterization and physical laws of acoustic propagation.The model can provide a more realistic underwater acoustic channel for the simulation analysis of turbo equalization algorithms than BELLHOP model which ignores channel variations.Second,system model and basic assumption of turbo equalization are provided.Two traditonal equalization algorithms,include Bahl-Cocke-Jelinek-Raviv(BCJR)algorithm and linear minimum mean square-error estimation(LMMSE)algorithm,are introduced.We mainly introduce several approximate methods for reducing the computional complexity of LMMSE algorithm and an improved real-number-notation LMMSE(RLMMSE)algorithm in quadrature modulation.Then,the basic theory of expectation propagation(EP)algorithm is deduced in this paper.After that,EP-Filter(EP-F)algorithm,an algorithm with high performance and low complexity,is introduced.To improve quadrature modulation performance,a hybrid expectation propagation(HEP)equilization algorithm is proposed based on the real number notation and its convergence is guaranteed by using the hybrid equalization strategy.Last,an improved proportionate normalized least mean squares(IPNLMS)algorithm is utilized to estimate above mentioned channel information efficiently.With the estimated information,different turbo equalization algorithms are simulated and analyzed.It is proved that EP algorithms perform better than traditional LMMSE algorithms.