中文摘要：

压缩感知理论应用于分布式认知网络中时，由于每个认知用户所处的信道环境差别很大，因此频谱感知的精度相差很大．为了提高感知的精度，提出了一种结合了分布式网络中的数据融合方法和压缩感知理论中的高效的数据处理方式的方法．首先，单个认知用户单独地运用压缩采样（CS）进行频谱的粗略感知，然后通过互信息的计算可以得到两两认知用户之间感知信息的差异，而差异大的两个认知用户之间会产生关联．认知用户的感知信息会按照这种关联进行共享．信息共享后，在每个认知用户端，基于贝叶斯推理的压缩感知恢复会重新进行来更新之前的感知结果．仿真结果表明，在感知精度与感知速率方面，算法性能均有改善．

英文摘要：

When compressive sensing is applied in cognitive radio network, spectrum sensing precision by every cognitive radio user is greatly different due to different channel environments between them. Consequently information fusion methods in network and the efficient data processing manner by compressive sensing can be combined to improve sensing precision. First, CS （compressive sampling） is performed independently by every cognitive radio user for rough sensing, and then the sensing information between different users is exchanged for their spatial diversity. Here, mutual information is taken as a measure to evaluate the sensing difference between two cognitive radio users, and those users with large difference are related. The sensing information of every cognitive radio user will be shared under this kind of relationship. After sensing information is shared, Bayesian inference for CS construction in every cognitive radio user is re-built to update the local sensing. The simulation results show that the proposed scheme has advantage both in sensing accuracy and in sensing speed over the conventional scheme.