The invention is characterized in that: generalized Legendre sequence or the discrete Fourier inverse transform is adopted in the steps as the training sequence;

• 所述方法包括：序列长度选择、序列选择、序列填充结构选择、信道估计算法，在上述步骤中使用广义勒让德序列或其离散傅立叶逆变换作为训练序列。