摘要
基于模式识别领域中的CCIPCA算法,该文给出了一种低运算量的在线Music算法。它无需估计协方差矩阵和对其进行特征值分解,信号子空间的估计与快拍数据的接收是同时进行的,而且只需存储当前的快拍数据,因此大大降低了存储量及运算量的要求;并针对上述算法在小快拍情况下性能较差的缺点,利用数据复用的方法有效提高了其估计性能。最后,计算机仿真验证了该文方法的有效性。
Based on CCIPCA algorithm in pattern recognition, a low complexity online Music method is presented firstly. It does not need to form the sample covariance matrix or compute its eigenvectors and the estimation of signal subspace begins after the first snap being received, this means the subspace estimation and the data receiving is simultaneous, and the current snap is the only data need to be stored. Then, the data is used repetitiously to improve subspace's estimation performance when few snap is available. Finally, experiments based simulated data demonstrate the efficiency of the presented algorithm.
出处
《电子与信息学报》
EI
CSCD
北大核心
2008年第11期2658-2661,共4页
Journal of Electronics & Information Technology