水声多途信道的维纳误差限信道质量量化分析方法

Quality quantitative analysis method for underwater acoustic multipath channel using Wiener error boundary

提出了一种定量分析不同多途结构的时不变水声信道对单载波水声通信系统性能影响程度的方法。该方法利用水声信道下信号线性均衡的残留均方误差与通信系统误比特率性能相关性强的特点,计算信号线性均衡的残留均方误差极限,并将其表示为只与信道冲激响应有关的变量,称之为维纳误差限(WEB),将此作为多途信道质量量化指标来评估多途信道对通信性能的影响。为验证方法性能,对比分析了大量海试实测水声信道下,线性均衡、判决反馈均衡以及Turbo均衡等单载波通信中3种常用信道均衡方法的误比特率与多种水声信道质量量化指标的相关性,利用Spearman相关系数对相关性进行量化分析,实验结果表明,相比于时延扩展、相干带宽、频谱衰落深度和信号多径比等传统信道多途特性量化指标,维纳误差限与均衡后误比特率的Spearman相关系数更高,平均相关系数可达0.71。利用维纳误差限可有效量化和分类不同质量的多途信道,定量体现不同水声信道多途结构对于通信性能的影响程度。

A channel quality estimating method is proposed,which can be used to quantitatively analyze the influence of timeinvariant channels with different multipath structures on the single-carrier underwater acoustic(UWA) communication performance.Since the correlation between the residual mean square error of the signal linear equalization under UWA channels and the bit error rates(BERs) of the communication system is strong,the residual mean square error boundary of the signal linear equalization is calculated and defined as a variable only related to the channel impulse response,namely Wiener error boundary(WEB),then it is utilized as a quantitative indicator of the qualities of multipath channels to evaluate the impact of multipath channels on the communication performance.To verify the performance of the method,the correlations between various quantitative indicators of sea trial UWA channel qualities and BERs after channel equalization are analyzed to compare the performance of corresponding indicators.The experimental results indicate that,compared with traditional indicators such as delay spread,coherence bandwidth,spectral null depth and signal multipath ratio(SMR),the Spearman correlation coefficient between WEB and BERs is higher,and the average correlation coefficient can reach 0.71.WEB can be used to quantify and classify multipath channels of different quality effectively,and reflect the impact of UWA channels with different multipath structures on communication performance quantitatively.