Blind Equalization Algorithms Using Higher-Order Statistics For Underwater Acoustic Channel

Nowadays, high data rate communication through band-limited underwater acoustic channel (UWAC) is of general interest in the field of underwater communication. Themain difficulty to achieve high data rate is the intersymbol interference (ISI). Periodic training sequence is essential for the conventional adaptive equalizers to combat the ISI, but it wastes plenty of bandwidth. Blind equalization (BE) needs no training sequence, which means higher data rate. Hence, it is significant to study the performance of blind equalizers for UWAC.The dissertation focuses on the performance of BE using higher-order statistics (HOS) implicitly and explicitly in accordance with the characteristic of UWAC. The research consists of several aspects as follows: The characteristics of UWAC is analyzed, including the multi-path effect, the Doppler shift effect, etc.. Then the mathematics' model of UWAC is given. The mathematics' model of Bussgang BE for UWAC is discussed. And in our research, the constant modulus algorithm (CMA) is emphasized, which is a special example of the Bussgang type. A kind of switch algorithm from CMA to decision directed (DD) algorithm is studied, and the selecting method of the decision circle, according to signal-to-noise ratio (SNR), is presented. To signaling with no constant modulus (such as 16QAM), the switch algorithm is able to achieve less residual Mean Square Error than that of CMA. Tricepstrum equalization algorithm (BTEA) and super-exponential (SE) algorithm based on block data estimation is studied, and these algorithms use HOS explicitly. Their performance, such as estimation variance and bias, is analyzed. A kind of sparse cross cumulant and sparse equalizer is proposed to simplify the SE algorithm, and the simulating results show efficient reduction in complication. From the SE algorithm, the relationship between the local minimum and the equalizer delay is demonstrated, and it is more accurate compared to the system delay relation formulation. The initialization method to achieve different equalizer delay local minimum is proposed for BTEA and SE. Comparison study using several UWAC with different zero locations is made to demonstrate the equivalent of different initialization method for least mean square (LMS) algorithm, BTEA, SE and CMA. On the basis ofthe above study, a kind of tap shift CMA is proposed, and it can achieve less residual MSE compared to center spike initialization. Comparison study between the constant modulus decision feedback equalizer (DFE) and several others is made on combating the frequency selective fading of UWAC, and their tracking performance to time varying is also investigated. A kind of DFE structure using SE iteration form is proposed here to accelerate its convergence rate, and its ability to adapt the nonstationary environment is also improved. In case where long-range communication is needed, spatial diversity blind equalizer (SDBE) is an effective way to improve SNR and suppress distortion. Joint carrier phase recovery with SDBE using second order digital phase lock loop is presented to track the Doppler shift, and a kind of quantization method is used to reduce the complexity of the multi-channel algorithm. Its performance is very similar to the original one. The BPSK data with statistics according with the demands of BE is achieved via pool test. Using these data, we demonstrate the efficiency of several BE algorithms, including comparison study and evaluation of their performance.This work was supported by the National Defense Science & Technology Key Lab of Underwater Acoustic Technology.