子空间盲信噪比估计算法研究

Study of Subspace Method on SNR Estimation

信号子空间维数估计是影响子空间信噪比估计算法性能的重要因素。针对利用最小描述长度（MDL）准则估计维数时,只能在有限信噪比范围内实现精确估计这一情况,采用噪声功率（NP）法与加权平均信息（WIC）准则以及归一化比值法对信号子空间维数进行估计,改善了原算法在低信噪比时估计性能变差的问题。仿真结果显示在加性高斯白噪声信道条件下,实际信噪比在-10~10 d B范围内时,噪声功率（NP）法以及加权平均信息（WIC）准则比原方法有显著的优势。

Estimation of the dimensions of signal subspace and noise space is an important factor in subspace methods on SNR estimation.MDL criterion leads a SNR-limited accurate estimation.With NP method or WIC method,the suhspace method will achieve a better performance in low SNR situation. Simulation results show that NP method and WIC method provide better performances in AWGN channel than traditional MDL methods when SNR ranges from -10dB to 10dB.