基于FPGA实现电话会议的研究

对电话会议进行了归纳分类。讨论了一种全双工预约会议。该会议采用FPGA对多方数字话音迭加分配,打破传统桥分器式系统对接口平衡阻抗的苛刻要求,彻底消除会议回声,避免会场声音振鸣(啸叫)。介绍了基于电路交换的电话会议工作原理,给出该会议的具体实现方法,并且举例说明具体操作步骤。     

N-S方程隐式分区并行计算

将Wu Z N和Zou H提出的一套适合三点块三对角隐式格式的定常和非定常可压无粘流的覆盖分区并行计算方法推广到N-S方程的计算.在二维定常亚音速、跨音速和非定常情况下都得到了较好的绝对并行效率.     

模式空间变换技术解析多组份光谱重叠体系及其在食用色素同时？…

采用模式空间变换技术使多组份光谱重叠体系中各组份得到有效地分离从而达到部分或所有组份定性及定量的目的。方法分辨效果好,定性定量结果可靠。无论光谱重叠程度如何,只要光谱间存在一定的差异即可得到较好的结果。将该方法用于人工合成食用色素多组份体系中各组份的同时测定,获得了较好的结果。     

红外系统信号恢复的三重相关重叠取样方法

提出了将三重相关和重叠取样技术用于红外系统信号波形恢复的方法,研究表明,对低输入峰值信噪比用该方法可以用效地从噪声中恢复信号的波形,文中描述了在重叠取样的基础上进行三重相关操作的步骤,比较了“三重相关重叠取样法”和传统的“扫描取样积分法”的掏噪性能的优劣,给出计算机模拟实验的结果。     

Cyclic-Auto-Correlation Based Timing Estimation Algorithm for Time-Frequency Overlapping Multi-Carrier Signals

In recent years, the time-frequency overlapping multi-carrier signal has been a novel and valuable topic in blind signal processing, especially in the non-cooperative receiving field. But there is little related research in public published papers. This paper proposes two timing estimation algorithms, which are non-data-aided and based on the cyclic auto-correlation function. In order to evaluate the performance of the proposed algorithms, the theoretical bound of the timing estimation is derived. According to the analyses and simulation results, the effectiveness of the proposed algorithms has been demonstrated. It shows that Method I has better performance than Method II. However, Method II does not need prior information, so it has a wider range of applications.     

L阶Markov信号的稀疏表示

在现实生活中,很多信号（比如语音信号）都具有有色性,即信号相邻采样点之间具有统计相关性,通常可采用L阶Markov过程进行较好的描述,然而已有的稀疏表示算法并没有充分考虑到这种统计特性。因此,针对L阶Markov信号,采用l（p≤1）-范数的广义平均值作为稀疏度量,并提出了基于重叠采样的稀疏表示算法。仿真结果表明,相比现有的线性规划稀疏表示方法、最短路径法和FOCUSS法,新算法的精度更高。     

A composite-structured-light 3D measurement method based on fringe parameter calibration

This paper proposes a novel method for reducing measurement error caused by spectrum overlapping in composite-structured-light 3D measurement systems. For a composite-structured-light 3D measurement system, spectrum overlapping causes parameters of each deformed phase-shifting fringe to change, and therefore leads to phase measurement errors. The proposed fringe parameter calibration method is based on the fact that variations in each deformed fringe's parameters are independent of height and reflectivity of the measured object. Three frames of composite grating are projected on the reference plane, and each carrier channel includes the information of three phase-shifting sinusoidal gratings used in Phase Measuring Profilometry (PMP). With the parameter calculation formulas of PMP, the parameters of fringes demodulated from the same carrier channel can be calculated, and therefore parameter relation coefficients between fringes demodulated from different carrier channels may be obtained. When an object is measured, these relation coefficients can be used to calibrate the parameters of the deformed phase-shifting fringes. A new 3D measurement mathematical model is established to reconstruct the shape of the object. Experimental data proved that the proposed method can effectively restrain the effect of spectrum overlapping and improve measurement accuracy by more than three times.     

基于拓扑势的重叠社区及社区间结构洞识别——兼论结构洞理论视角下网络的脆弱性

社会网络和复杂网络上的社区识别已经成为当前研究的热点和前沿课题.针对目前社区识别方法不能兼具较低时间复杂度、无须专家知识或先验知识和允许存在重叠节点的不足,提出了基于拓扑势理论的重叠社区识别方法.通过提出的重叠节点社区归属不确定性测度,该方法同时实现了社区间结构洞的识别.实验验证了该方法的有效性.另外,文章在理论证明的基础上提出了影响因子优化算法;论证了结构洞理论视角下网络的脆弱性.     

一种低存储量需求的高效LDPC译码方法

提出了一种高效的低密度奇偶校验码（LDPC）译码方法,相对于传统译码方法,仅需增加少量存储量便能获得接近2 倍的吞吐率增益.本文将修正的最小和算法与分组双向消息传递（STMP）译码算法相结合,提出了基于最小和的分组双向消息传递算法（MS-STMP）,并给出了相应的迭代交叠方案.随后讨论了交叠过程中2 种运算单元对存储器的访问.最后以中国地面数字电视传输（DTTB）标准中使用的一组LDPC 码为例,计算了MS-STMP 算法相对于传统双向消息传递（TPMP）算法的吞吐率增益和额外增加的存储量.计算结果表明,MS-STMP 算法平均增加44%的存储量,而将吞吐率平均提高到1.85 倍,相对于交叠（OMP）算法有明显的优势.