The Application Research Of Sparse Bayesian Learning In Underwater Acoustic Communication Doppler Estimation

In high speed underwater acoustic communication system, when signal transform from sparse underwater acoustic channel, the signal usually distorts because of multipath fading, time delay and frequency-selective fading. So far, some studies have found that the orthogonal frequency division multiplex could solve the problem of frequency-selective fading. Also the multi-input multi-output could solve the problem of multipath fading, and transform the multipath problem to profitable operation. Therefore unite MIMO technique and OFDM's these two techniques the consideration becomes a kind of very natural viewpoint. But in mobile acoustic communication system, there is relative speed between sender and receiver lead to Doppler effect. The Doppler effect concrete manifestation for extension of received signal, causing the signal distorted, this may led to get wrong information when processing the receive signal. So in mobile underwater communication system, it is necessary to estimate the Doppler factor.The compressive sensing aimed at sparse signal, sample a signal and compresses at the same time, so only need a few samples can reconstruct spare signal. Because of the inherent sparse in underwater channel, we can use the compressive sensing theory on underwater acoustic signal. So in this paper, we can transform the Doppler factor estimate to underwater acoustic signal processing, and use this method to solve the problem. The reconstruction algorithm we choose the Sparse Bayesian Learning Algorithm, when it is running, iteration self-learning controller need't identify system parameters but adjust income to develop the performance of system, this advantage perfect fit with the underwater acoustic channel unknown sparsity characteristic. So we study on underwater acoustic Doppler base on SBL algorithm.We first introduce the underwater acoustic OFDM technology and so on, and based on the sparsity of underwater acoustic channel we proposed the new Doppler estimate method. This method use the SBL to refactoring the received comb-type pilot to get the impulse response of channel, and combining the autocorrelation function to calculation the Doppler factor. The simulation analyses prove that the method has higher estimation accuracy and depth of theoretical study.Subsequently, we analyze the MIMO-OFDM system, and put forward a new method of solving the Doppler problems existing in the system. In this system we need to consider the impact on sparsiey of MIMO acoustic channel, that while the signal transform through underwater acoustic channel, there are only a few signal can pass the channel to the receiver end, and there are different time-delay factors and frequency offset. In this case, we solve this problem by design a joined pilot and synchronous code use the SBL to estimate the system Doppler factor. The simulations show that, this method not only can estimate the different sub-channel Doppler factor but also has anti-noise property.