OFDM系统基于二维核回归算法的衰落信道估计

提出了一种低复杂度的基于二维核回归平滑的时间频率双衰落信道估计算法.该方法首先采用LS算法估计导频处的信道增益并利用分段线性插值方法得到数据子载波处的信道增益,然后分别利用两个级联的一维核回归平滑器在时域和频域对信道矩阵进行二维平滑,以消除由信道噪声、ICI和插值误差等引入的估计误差.对插值和平滑分别进行优化处理,因此具有计算复杂度低、性能优良等优点.数值仿真表明:该算法性能较LS算法和1维DFT算法有较大的改进;与1维LMMSE算法相比性能相差很小,在低信噪比时甚至超过了后者.

2D Kernel Regression Smoothing Method-Based Channel Estimation for OFDM Systems

A low complexity 2D kernel regression smoothing method, which employed two separate 1D smoothers in frequency-domain and time - domain accordingly, was proposed to eliminate the estimation error introduced by channel noise, Inter Carrier Interference (ICI) and interpolation error. The proposed method processes the smoothing and interpolation tasks in a separate manner, so the optimization process can be taken separately too. The algorithm has the advantages of low computational complexity and high performance. Numerical simulations show the proposed algorithm outperforms the LS method and 1D DFT method greatly and has performance close to 1D LMMSE algorithm.