采用定点运算FTF算法的自适应噪声抵消器的计算机仿真

Computer Simulation of Adaptive Noise Canceller Using Fixed-point FTF Algorith

本文讨论用快速横向滤波器(FTF)算法实现自适应噪声抵消(ANC),着重研究采用16位定点运算的情况.但是FTF算法具有众所周知的数值性能不稳定性,该问题在用有限字长定点实现时变得尤为明显.加外因为FTF算法包含除法运算,采用定点运算后,由于有限精度的影响,除法中的分母和转换因子γ有时会衰减为零,从而导致算法不能正常工作.本文针对这些问题讨论了相应的解决方法.另外,本文采用定标的方法避免了FTF算法中的某个正变量变为负值,从而保证了算法的稳定.最后本文给出了计算机模拟的结果.

In this paper, the realization of adaptive noise cancellation using fast transveral filter (FTF) algorithm is discussed, and the case the uses a 16 bit fixed-point arithmetic is emphatieally studied. Numerical unstability of FTF algorithm becomes conspicuous especially when the algorithm is implemented by the finite-word-length fixed-point arithmetic. Because he FTF algorithm contains division operations, the denominator and the conversion factor may attenuate to zero due to the effect of finite percision, and this will make the algorithm not to work normally. Relevant methods for solving these problems are discussed. Furthermore, demarction is adopted to prevent a certain positive variable in the algorithm becoming negitive. Thus, the stability of the algorithm is pledged. Finally, the computer simulation results are given.