1: \begin{abstract}

2: We propose a novel M-estimate conjugate gradient (CG) algorithm,

3: termed Tukey's biweight M-estimate CG (TbMCG), for system

4: identification in impulsive noise environments. In particular, the

5: TbMCG algorithm can achieve a faster convergence while retaining a

6: reduced computational complexity as compared to the recursive

7: least-squares (RLS) algorithm. Specifically, the Tukey's biweight

8: M-estimate incorporates a constraint into the CG filter to tackle

9: impulsive noise environments. Moreover, the convergence behavior of

10: the TbMCG algorithm is analyzed. Simulation results confirm the

11: excellent performance of the proposed TbMCG algorithm for system

12: identification and active noise control applications.

13: \end{abstract}