遗传优化神经网络的水声信道盲均衡

不需要训练序列的盲均衡技术可以有效地节省水声通信带宽，消除码间干扰，提高水声通信效率和质量。以前馈神经网络（FNN）作为盲均衡器，既适用于最小相位信道，也适用于非最小相位信道，包括非线性信道，但是前馈神经网络在实际的应用中其网络拓扑结构的选取和初始权重的确定缺乏理论依据，且其训练主要依靠BP算法，存在收敛速度慢、容易陷入局部极值及“过学习”的问题。为此，本文提出了一种遗传优化神经网络的水声信道盲均衡算法（GA—BP），对前馈神经网络拓扑结构和网络权重同时优化，有效地克服了传统前馈神经网络盲均衡的缺陷，提高了前馈神经网络盲均衡的泛化性能并加强了跟踪时变信道的能力和对信道突变的适应能力。水池试验结果证明了文中提出的遗传优化神经网络水声信道盲均衡算法的有效性，与直接前馈神经网络盲均衡相比较，均衡性能明显得到了提高。     

水声通信中基于信道辨识的盲turbo均衡方法

水声信道多途效应明显,造成接收信号存在严重的码间干扰(ISI,Intersymbol interference)。基于最小均方误差(MMSE,Minimum mean square error)准则的turbo均衡器级联了均衡和信道译码,能够有效去除ISI,并获得优良的性能。由于水声信道的时变性,传统MMSE-turbo均衡需要周期性的训练序列,以实现连续可靠的通信。训练序列虽然提高了通信的可靠性,但降低了信息的有效传输速率。因此,为提高通信效率,本文提出了一种盲turbo均衡方法,该方法通过引入新的盲信道辨识器来同时获得信道估计响应和已去除部分ISI的初步均衡输出信号,并为turbo均衡提供初始的响应参数和比特软信息。与水声通信中应用较多的盲判决反馈均衡器(DFE,Decision feedback equalizer)相比,海上实验结果证明本文提出的盲turbo均衡方法抗信道多途衰落的能力较强,并且与传统MMSE-turbo均衡相比无需训练序列,因此提高了信息的有效传输速率。     

基于集成综合评价法和聚类分析法的综合评价方法

针对传统的主成分分析法在数据无量纲化中"中心标准化"的缺点,提出了改进的主成分分析方法,并建立了一种基于集成综合评价法和聚类分析法的新的综合评价方法,通过黑龙江省各城市综合竞争力的统计分析结果说明该方法能有效的得出合理结果.     

基于量子粒子群优化的正交小波加权多模盲均衡算法

针对常数模盲均衡算法(CMA)均衡高阶正交振幅调制信号(QAM)存在收敛速度慢、稳态误差大的缺点, 提出了基于量子粒子群优化的正交小波加权多模盲均衡算法(QPSO-WTWMMA). 该算法根据高阶QAM信号星座图分布特点, 将量子粒子群优化算法(QPSO) 和正交小波变换融入于加权多模盲均衡算法(WMMA)中. 因而, 利用QPSO对均衡器权向量进行了优化, 利用正交小波变换降低了输入信号的自相关性, 利用WMMA选择了合适的误差模型匹配QAM星座图. 理论分析及水声信道仿真结果表明, QPSO-WTWMMA算法可以获得更快的收敛速度和更低的稳态误差, 在水声通信中具有重要的参考价值.     

水声信道线性均衡的研究

本文在对LMS和稀疏权自适应均衡算法进行分析的基础上，将其引入到水声数据传输中，利用LMS算法对时不变信道和慢变信道进行了仿真实验；利用稀疏权自适应均衡算法对深海信道进行了仿真实验，并获得了仿真结果。     

UMFSK水声通信技术研究

在MFSK基础上,本文提出了UMFSK信道编码方法。采用窄带调频和跳频调制技术,故在一定程度上抑制了由多途效应引起的码内/码间干扰;提出了基于UMFSK的实时信道标校技术,可对信道进行稳键均衡,抗信道多途引起的码内干扰。进行了湖试,取得了试验数据并进行了数据分析。分析表明：UMFSK可抑制码元相消干涉引起的码内干扰,其效果随码元调频带宽的增大而变好;信道标校技术可对信道进行稳键均衡,并由于结构简单,可做到实时实现。基于UMFSK的误码率为10^-3（通信带宽2kHz,速率100-150bps)。     

语音通信降噪研究

语音通信系统中,语音通过信道传输将不可避免地引入码间串扰和信号畸变,同时受到噪声污染。本文在分析自适应盲均衡算法CMA(constant modulus algorithm）和改进盲均衡算法的基础上,考虑到自适应盲均衡技术在语音噪声控制方面能力有限,将自适应盲均衡技术与小波包掩蔽阈值降噪算法联合使用,形成一种基带语音增强新方法。仿真试验结果显示自适应盲均衡技术可以使星座图变得清晰而紧凑,有效减小误码率。研究证实该方法在语音信号ISI和畸变严重情况下,在白噪及有色噪声不同的噪声环境中都具有稳定的降噪能力,消噪同时可获得汉语普通话良好的听觉效果。     

Widely linear equalization for IQ imbalance contaminated multicarrier systems

In a multicarrier modulated (MCM) wireless communication system, the received bandpass signal is downconverted to the baseband signal for ease of receive processing. During downconversion, a slight mismatch in the amplitude and phase of locally generated sinusoidal signal in the In-Phase and Quadrature (IQ) branches of downconverter results in IQ imbalance of the baseband signal. The IQ imbalance leads to mirror interference between subcarriers of an MCM system and degrades the performance. To compensate for the effects of IQ imbalance, this paper presents widely linear minimum mean square error (WL-MMSE) based equalization technique and compares it with the traditionally used linear minimum mean square error (L-MMSE) based equalization technique. A performance comparison between WL-MMSE and L-MMSE based estimation techniques in terms of mean square error (MSE) and bit error rate (BER) suggests that the former outperforms the latter in the presence of IQ imbalance.     

Energy flux equalization: An open boundary condition for non-linear free surface flows

In the simulation of non-linear free surface flows in a finite domain a major concern is with the radiation condition to be applied at the ‘open’ boundary. No general theoretical radiation conditions are known to exist. In this paper a new open boundary condition is formulated based on energy flux equalization between the non-linear inner domain and a linear outer domain. The non-linear flow in the inner domain is solved numerically using a semi-Lagrangian procedure. The energy flux arriving at the open boundary is removed using a new open boundary condition which acts as a linear wave absorber. For the cases studied the performance of this boundary condition is found to be quite good.     

PMMA adsorption over previously adsorbed PS studied by polarized FTIR-ATR

Polarized infrared spectroscopy in attenuated total reflection was used to investigate the adsorption of PMMA (polymethylmethacrylate) onto PS (polystyrene) that was previously adsorbed onto oxidized silicon from dilute solution in carbon tetrachloride at 30°C. The carbonyl group of PMMA forms hydrogen bonds with surface silanol groups, giving segment-surface interaction energy of 4 kT as against 1 kT for PS (k is the Boltzmann constant, T the absolute temperature). The formation of hydrogen-bonding by PMMA was unaffected, in rate or amount, by preadsorbed PS, but a lesser total amount of PMMA adsorbed onto PS, resulting in a higher average bound fraction. The histogram of the mass adsorbed, attributable to subpopulations of chains with different bound fractions, was inferred by subtracting infrared spectra acquired at successive times. Whereas this histogram was broad and bimodal for adsorption onto bare surface, it was narrower and unimodal for adsorption onto preadsorbed PS. © 1995 John Wiley & Sons, Inc.