中国邮电高校学报(英文) ›› 2014, Vol. 21 ›› Issue (3): 106-112.doi: 10.1016/S1005-8885(14)60307-1

• Artificial Intelligence • 上一篇    

DOA estimation in unknown colored noise using covariance differencing and sparse signal recovery

田野, 孙晓颖, 秦宇镝   

  1. College of Communication Engineering, Jilin University, Changchun 130022, China
  • 出版日期:2014-06-30
  • 通讯作者: TIAN Ye E-mail:tianfield@126.com
  • 基金资助:

    This work was supported by the National Natural Science Foundation of China (61171137).

DOA estimation in unknown colored noise using covariance differencing and sparse signal recovery

田野, 孙晓颖, 秦宇镝   

  1. College of Communication Engineering, Jilin University, Changchun 130022, China
  • Online:2014-06-30
  • Supported by:

    This work was supported by the National Natural Science Foundation of China (61171137).

摘要:

A direction-of-arrival (DOA) estimation algorithm is presented based on covariance differencing and sparse signal recovery, in which the desired signal is embedded in noise with unknown covariance. The key point of the algorithm is to eliminate the noise component by forming the difference of original and transformed covariance matrix, as well as cast the DOA estimation considered as a sparse signal recovery problem. Concerning accuracy and complexity of estimation, the authors take a vectorization operation on difference matrix, and further enforce sparsity by reweighted ?1-norm penalty. We utilize data-validation to select the regularization parameter properly. Meanwhile, a kind of symmetric grid division and refinement strategy is introduced to make the proposed algorithm effective and also to mitigate the effects of limiting estimates to a grid of spatial locations. Compared with the covariance-differencing-based multiple signal classification (MUSIC) method, the proposed is of salient features, including increased resolution, improved robustness to colored noise, distinguishing the false peaks easily, but with no requiring of prior knowledge of the number of sources.

关键词:

direction-of-arrival, covariance differencing, sparse signal recovery, reweighted ?1-norm penalty, unknown covariance

Abstract:

A direction-of-arrival (DOA) estimation algorithm is presented based on covariance differencing and sparse signal recovery, in which the desired signal is embedded in noise with unknown covariance. The key point of the algorithm is to eliminate the noise component by forming the difference of original and transformed covariance matrix, as well as cast the DOA estimation considered as a sparse signal recovery problem. Concerning accuracy and complexity of estimation, the authors take a vectorization operation on difference matrix, and further enforce sparsity by reweighted ?1-norm penalty. We utilize data-validation to select the regularization parameter properly. Meanwhile, a kind of symmetric grid division and refinement strategy is introduced to make the proposed algorithm effective and also to mitigate the effects of limiting estimates to a grid of spatial locations. Compared with the covariance-differencing-based multiple signal classification (MUSIC) method, the proposed is of salient features, including increased resolution, improved robustness to colored noise, distinguishing the false peaks easily, but with no requiring of prior knowledge of the number of sources.

Key words:

direction-of-arrival, covariance differencing, sparse signal recovery, reweighted ?1-norm penalty, unknown covariance