论文部分内容阅读
本文提出了一种采用ADM(自适应增量调制)的DFT新结构,该结构硬件实现简单,无乘法及成本低,与基于LDM(线性增量调制的)的DFT结构相比,大大提高了信号功率的动态范围。对于某个确定的采样率系数K=f_s/2f_m(f_s是采样率,f_m是信号的带宽),只要使得量化噪声(散粒噪声和斜率过载噪声)达到最小,便能得到最佳步长。由于ADM的噪声难于精确计算,我们可以用它的平均步长来代替LDM的步长,因而用LDM的信噪比来估计ADM的信噪比,由此,推导了一个通过LDM的步长来计算ADM的步长的公式,这样就可以得到最佳步长。对于带限高斯信号进行计算机模拟的结果表明其性能接近于理论分析。与基于LDM的DFT硬件实现相似,在浮点运算中,除了ROM,加法器和其它辅助电路外,增加了一个移位寄存器,以便随着步长的变化来控制系数。
In this paper, we propose a new DFT structure using ADM (Adaptive Delta Modulation), which has the advantages of simple hardware implementation, no multiplication and low cost. Compared with the DFT structure based on LDM (Linear Incremental Modulation) Dynamic range of signal power. For a certain sampling rate factor K = f_s / 2f_m (f_s is the sampling rate and f_m is the bandwidth of the signal), the best step size can be obtained as long as the quantization noise (shot noise and slope overload noise) is minimized. Since the ADM noise is difficult to calculate accurately, we can use its average step size instead of the LDM step size, so we use the signal to noise ratio of the LDM to estimate the SNR of the ADM. From this we derive a step size Calculate the ADM step size formula, so you can get the best step. The results of computer simulation of the band-limited Gaussian signal show that its performance is close to the theoretical analysis. Similar to LDM-based DFT hardware implementations, in floating-point operations, a shift register is added in addition to ROMs, adders, and other auxiliary circuits to control the coefficients as the step size changes.