Research On Underwater Acoustic OFDM Channel Estimation And Tracking Based On Dynamic Compressive Sensing

OFDM technology has been widely used in bandwidth-limited underwater acoustic communication because of its high frequency band utilization and adjustable parameters characteristic.Channel estimation technology is a key technology in OFDM systems.Real-time and accurate channel estimation is of great significance for data demodulation and recovery at the receiver.Compared with wireless communication,underwater acoustic signals are more affected by multi-path delay spread and channel time-varying characteristics.Although the OFDM symbol duration can be made smaller than the channel coherence time by changing the communication parameter setting,the influence of the time varying characteristic of the channel on the communication signal can be reduced.However,this will also amplify the effects of multipath fading.Traditional channel estimation methods must increase the number of pilots to maintain the same performance.However,with the deepening of research,the multipath distribution of underwater acoustic channels has been proved to be sparse.By using compressed sensing theory,underwater acoustic channel estimation problems can be converted into sparse signal reconstruction problems,and channel estimation performance can be effectively improved without increasing the number of pilots.However,existing methods for underwater acoustic channel estimation based on compressed sensing mostly adopt symbol-by-symbol methods,which often have the following two problems: First,current compressed sensing reconstruction algorithms have high complexity and difficulty in hardware implementation.Therefore,they are difficult to apply to channel tracking and other applications that require high real-time performance.Second,the vast majority of existing studies only consider the intrinsic sparsity of underwater acoustic channels in static time,and consider the situation under dynamic channels less frequently,ignoring the correlation between sparse multipath channels with adjacent time before and after,failing to explore the full potential of compressed sensing theory.This dissertation focuses on the above two problems,and studies the dynamic compressive sensing theory for dealing with time-varying sparse signals,and correlates the dynamic changes of underwater acoustic channels with time-varying sparse signals,and uses OFDM underwater acoustic communication as an application background to develop two kinds of sparse signal models with time-varying channel characteristics.Based on this,separately addressing the situation in which the above two problems lie,under the assumption that the underwater acoustic channel changes symbol by symbol,the temporalcorrelation characteristics of multipath structures in underwater acoustic sparse channels are fully exploited,and two improved compressed sensing algorithms are proposed.Theoretical analysis and simulation results show that the two proposed algorithms are well adapted to the above two scenarios.The innovations of the two proposed algorithms are as follows:A)Improved SOMP algorithm: Research on distributed compressive sensing theory aims to convert the problem of channel impulse response estimation in consecutive time slots into the problem of joint restoration of multiple sparse signals in the space that distributed compressed sensing can handle.According to the slow time-varying characteristics of underwater acoustic channels,the proposed algorithm is an improved algorithm based on SOMP algorithm.By selecting the JSM1 model to jointly recover the channel and increase the detection of independent taps in each time slot,a higher estimation performance is achieved.B)Dynamic OMP algorithm: Utilizing the temporal correlation of the impulse response of underwater acoustic channels,the channel variation is simulated through the AR process,and a dynamic sparse observation model is established.The algorithm only performs a complete OMP channel estimation to obtain the channel support set at the initial time,and then tracks the channel by continuously tracking the change of the channel support set at the previous time,thus effectively reducing the complexity of the algorithm and can be better applied to the channel tracking.