基于训练序列的水声信道多普勒频移估计

Doppler Frequency Shift Estimation of the Underwater Acoustic Channel Using Training Symbols

由于水声通信媒介(声波)速度远远小于陆地通信媒介(无线电波)速度,水声通信多普勒频移比陆地无线通信多普勒频移要大几个数量级,因此多普勒频移估计是水声通信的一个重要问题。文中提出了一种水声通信系统中基于训练序列的多普勒频移估计方法。在接收端先由已知训练符号构造无噪声干扰下的理论接收信号;再根据理论接收信号与实际接收信号之间的误差平方和最小,得到含有信道衰减因子A和多普勒因子Δ的双变量目标函数,根据此目标函数的曲面图以及对其的分析,发现A的不同取值基本不影响Δ的寻优,从而可将双变量目标函数变成单变量目标函数;然后利用全局优化理论中求解有限区间内多峰值函数最值的区间斜率算法,求解多普勒因子的最小二乘估计值,从而得到多普勒频移的估计值。文中还分析了区间斜率算法的收敛性,且仿真结果显示所提算法优于现有算法。

The velocity of sound wave in the water is far less than that of radio wave, and Doppler frequency shift in underwater acoustic communication is several orders of magnitude greater than that in wireless communication on land. Therefore, Doppler frequency shift estimation is an important problem in underwater communication. A Doppler frequency shift estimation method is presented in this paper for underwater acoustic communication system based on training sequence. In the receiving port, the theoretical received sequence is composed without noise interference and on the basis of known training symbols. Then the objective function of two variables(channel fading factor A and Doppler factor Δ) is derived in line with the Least Square Principle which minimizes the error between the theoretical and actual received signals. According to the surface diagram and analysis of the objective function, it is discovered that different values of A will not affect the optimization of Δ, meaning it is feasible to translate the two-variable function into a single-variable objective function. The least squares