论文部分内容阅读
本文讨论了使级联的模乘和标度运算从三次减少到两次的剩余数制(RNS)自适应最小均方算法;分析和计算了三个简单实用的RNS自适应噪声对消器(自适应系统):1.两模RNS自适应系统,其速度比有同等硬件量和精度的二进制自适应噪声对消器高一倍;2.极经济地实现无内部运算误差的两模RNS权调整三模RNS加权对消的RNS自适应系统;3.用多微处理器实现的多素数模RNS自适应系统。剩余算术不但可以显著改善有固定加权系数的滤波器的性能价格比,而且如本文所例证的那样,也能大大提高加权系数可变的自适应数字滤波器的性能价格比。计算机模拟与理论分析一致。
In this paper, the RNS adaptive minimum mean square algorithm that reduces the modular multiplication and scaling operations of cascades from three to two is discussed. Three simple and practical RNS adaptive noise cancellers are analyzed and calculated Adaptive system): 1. Two-mode RNS adaptive system, which is twice as fast as the binary adaptive noise canceller with the same amount of hardware and precision; 2. Very economical implementation of two-mode RNS right without internal computational errors Adjust the RNS adaptive system of three-mode RNS weighted cancellation; 3. multi-prime modulus RNS adaptive system realized by multi-microprocessor. Remaining arithmetic not only significantly improves the performance-to-price ratio of a filter with a fixed weighting factor, but also, as exemplified herein, greatly increases the cost-effectiveness of adaptive digital filters with variable weighting coefficients. Computer simulation and theoretical analysis.