基于多通道的高速并行采样信号重建算法研究

发布时间：2018-04-28 16:58

本文选题：非均匀采样 + 信号重建　；参考：《电子科技大学》2017年硕士论文

【摘要】：随着现代电子信息技术的快速发展,信号高速采样的应用越来越广泛。例如在宽带测试系统、雷达、遥测遥感、图像处理、宽带模拟IC等方面都需要用到高速数据采集系统。但根据奈奎斯特原理,要想完全无损地重建信号,采样频率至少是信号频率的两倍以上。实际上,在高速信号领域,单片的ADC并没有达到如此高的采样率。于是在20世纪80年代,一些研究学者提出的时间交织采样结构有效地解决了这个问题,多通道并行采样系统就是采用多片低速ADC形成的时间交织结构对信号进行采样,并在后端进行组合,从而实现了对信号的高速采样。实际的TIADC(Time-interleaved Analog-to-digital Converter)系统由于各个通道间的失配误差(偏置误差、增益误差、时间误差)的存在,使得采样后的信号形成了非均匀采样波形,并引入了杂乱频谱。因此要想完全重建信号,首先要对信号之间的失配误差进行估计,然后对这些误差进行有效地校准。本文主要是从通道误差估计和误差校准两个方面来进行研究,考虑到现有的方法多少都有一些不足,因此本文对失配误差中的估计和校正算法都进行了改进,进行了更加深入的研究。(1)对失配误差的估计进行了改进。经典的正弦拟合法在进行误差估计的时候,大多需要事先知道信号的频率。而在大多数应用中,信号的频率是无法准确得知的。因此,本文对失配误差的估计进行了“两步分解”。首先,采用基于相位差的快速傅里叶变换对信号的频率进行了精确的估计,再采用经典的正弦拟合法对信号进行失配误差的估计。将这两种算法结合,这样就可以在无需知道信号频率的情况下,也可以对信号进行精准的失配误差估计。(2)对时基误差的校准采用了三次样条插值法,并且根据采样后信号形成的周期特性,对运算进行了简化。在采样只存在时基误差的情况,对三次样条插值法进行信号重建做了详细的阐述说明,利用形成的周期非均匀采样的周期性对三次样条插值法进行了改进,降低了运算的复杂度。对分数延迟滤波器进行信号重建做了理论推导,并对比三次样条插值法和分数延迟滤波器法。为了证明改进后算法的有效性,本文采取了matlab仿真验证。仿真结果表明,使用两种算法结合,对于通道失配误差的估计是可行的,并且误差估计精度较高。而在信号重建方面,改进后的三次样条插值法在无杂散动态范围和信噪比等关键指标方面得到明显提升。
[Abstract]:With the rapid development of modern electronic information technology, the application of high-speed sampling signal is more and more extensive. For example, high-speed data acquisition systems are needed in wideband test system, radar, remote sensing, image processing, broadband analog IC and so on. But according to the Nyquist principle, the sampling frequency is at least twice as high as the signal frequency if the signal is to be reconstructed completely without loss. In fact, in the field of high-speed signals, the single-chip ADC does not achieve such a high sampling rate. Therefore, in the 1980s, some researchers put forward a time-interleaved sampling structure to effectively solve this problem. The multi-channel parallel sampling system uses the time-interleaved structure formed by multi-chip low-speed ADC to sample the signal. And the combination in the back end, so as to achieve the high-speed sampling of the signal. Because of the mismatch error (bias error, gain error, time error) between the various channels, the sampled signal forms the non-uniform sampling waveform and introduces the clutter spectrum in the actual TIADC(Time-interleaved Analog-to-digital conversion system. Therefore, in order to reconstruct the signal completely, we must estimate the mismatch error between the signals, and then calibrate these errors effectively. In this paper, the channel error estimation and error calibration are mainly studied. Considering that there are some shortcomings in the existing methods, the estimation and correction algorithms of mismatch errors are improved in this paper. The estimation of mismatch error is improved. The classical sinusoidal fitting method needs to know the frequency of the signal in advance in error estimation. In most applications, the frequency of the signal is unknown. Therefore, the estimation of mismatch error is decomposed by two steps. Firstly, the fast Fourier transform based on phase difference is used to estimate the frequency of the signal accurately, and then the classical sinusoidal fitting method is used to estimate the mismatch error of the signal. By combining these two algorithms, we can use cubic spline interpolation method to calibrate the time-base error without knowing the frequency of the signal, and to estimate the mismatch error of the signal accurately. The operation is simplified according to the periodic characteristic of the sampled signal. When sampling has only time base error, the cubic spline interpolation method is explained in detail, and the cubic spline interpolation method is improved by using the periodicity of periodic nonuniform sampling. The computational complexity is reduced. The signal reconstruction of fractional delay filter is theoretically deduced and compared with cubic spline interpolation method and fractional delay filter method. In order to prove the effectiveness of the improved algorithm, matlab simulation is adopted in this paper. The simulation results show that it is feasible to estimate the channel mismatch error with the combination of the two algorithms, and the accuracy of the error estimation is high. In the aspect of signal reconstruction, the improved cubic spline interpolation method has been improved obviously in terms of the non-spurious dynamic range and signal-to-noise ratio (SNR).
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN911.7

【参考文献】