利用组稀疏特性的宽带压缩频谱感知

基于认知无线电的动态频谱接入需对宽带信道进行频谱感知，而越来越高的采样速率日益成为宽带频谱感知的瓶颈。压缩感知作为一种新的信号获取技术为亚奈奎斯特采样速率下的宽带频谱感知提供了一种可行方案。在相关应用场景中，如果能够挖掘相关先验信息并在重构算法中整合这些信息，将大幅提高压缩感知的性能。本文基于压缩感知技术，利用信道的划分信息及宽带信号的组稀疏特性，提出了一种组稀疏贪婪算法GOMP。该方法在成熟的贪婪算法基础上，利用子信道内多频点的组测量信息，根据组测量的概率分布特性来识别宽带信道的活动子信道。这种组测量识别方式使算法能以较少的观测数据实现对宽带信道的快速准确感知，极大地降低了宽带频谱感知所需的采样速率。实验结果表明:该算法比传统的OMP算法及BP算法不仅具有更好的重构效果及频谱检测性能，而且具有更好的压缩性能及实时性能。 

Compressive wideband spectrum sensing using group sparsity

Wideband dynamic spectrum access based on cognitive radio has encountered bottleneck in the required speed of traditional sampling techniques. Compressive sensing (CS) emerged as an alternative candidate for wideband spectrum sensing at sub-Nyquist Sampling rates. Furthermore, exploiting and incorporating the prior information of related signals will signi?cantly reduce the sampling rate required by CS. Considering the sub-channel locations are known in advance, this paper proposes a new greedy algorithm, called Group-sparsity Orthogonal Matching Pursuit (GOMP), to reconstruct the wideband spectrum estimation by exploiting the characteristic of group sparsity corresponding to the channel distribution. The proposed algorithm based on the principle of the sophisticated greedy algorithm Orthogonal Matching Pursuit (OMP) incorporates a multipoint measurement of sub-channels to identify the signal support where active primary users are located. The multipoint measurement strategy converts the components of sub-channels into a L2-regularization, and then introduces the statistical characteristic of the regularization to identify the active sub-channels. This endows the algorithm with accuracy and robustness for signal support identification and spectrum reconstruction. A comparison experiment shows the proposed algorithm outperforms traditional OMP algorithm and the famous Basis Pursuit (BP) algorithm in terms of reconstruction errors and detection accuracy with fewer measurements. Moreover, it is faster than the two previous algorithms.