Research On Equalization Algorithm In Time-varing Multipath Underwater Acoustic Channels

In recent years, underwater acoustic communication(UWAC) plays an important role inthe field of military and civilian, and the importance of the role will become more and moreobvious as time goes by. The underwater acoustic channel has serious multipath, timevariation, and limited communication bandwidth. Comparing to traditional frequency divisionmultiplexing technology, Orthogonal frequency division multiplexing(OFDM) technologyhas higher bandwidth utilization rate because of using the orthogonality between subcarriers.By adding cyclic prefix, OFDM can eliminate the inter symbol interference(ISI) caused bymultipath effects. Because of time-varying, underwater acoustic channel will have badinfluences including doppler expansion, destroying the orthogonality between subcarriers,resulting in inter carrier interference(ICI) and reducing the performance of UWAC system.The purpose of channel equalization is to compensate the loss of signal in the amplitude,frequency and phase, eliminate the effects of ISI and ICI during the process of transmission.First of all, this paper introduces the physical characteristics of underwater acousticchannel, the basic principle of OFDM and the block diagram of underwater acoustic OFDMcommunication system. And then analyzes the causes of the ICI from the offset carrierfrequency, the offset sampling frequency and the effect of doppler spread. Research on theequalization algorithm for time-varying multipath channel, such as zero forcing(ZF)equalization algorithm, minimum mean square error(MMSE) equalization algorithm andone-tap(OT) equalization algorithm. And analyzes the performance of the algorithm throughexperimental simulation, the simulation channel consisted of the Rayleigh fading channel andunderwater acoustic channels generating by two kinds of underwater channel simulator.Channel equalization algorithm includes the inversion channel frequency-domain matrix(CFM) operation, however, the dimension of CFM is large and the time complexity of thedirect inverse operation is very high. Because ICI mainly comes from the the interference ofneighboring carriers instead of far carriers, the CFM can be approximately expressed by abanded matrix. At the end of this paper, a method was analyzed to reduce computationalcomplexity of channel equalization algorithm by decomposing band matrix using LU method.As for the serial block equalization algorithm, its complexity can also be decreased by theblock iterative inversion. Simulation results show that the reduction of banded matrixinversion can significantly decrease the computational complexity of several variationschannel equalization algorithm and ensure the performance of the equalization algorithm.