We propose a novel framework for enabling scalable database-driven dynamic spectrum access and sharing of heterogeneous wideband spectrum. The proposed framework consists of two complementary approaches that exploit the merits of compressive sensing theory, low-rank matrix theory, and user cooperation to build an accurate heterogeneous wideband spectrum map by overcoming the time-variability of the number of occupied bands, the need for a high number of measurements per sensing node (SN), the inherent wireless channels’ impairments, and the high reporting network overhead. First, exploiting the fact that close-by SNs have a highly correlated spectrum observation, we leverage distributed compressive sensing to enable cooperative heterogeneous wideband spectrum sensing only from a small number of measurements per each SN. Second, to reduce the network overhead due to the high width of the spectrum of interest, we propose a two-step approach that performs spectrum occupancy recovery using the local low-rank property of occupancy sub-matrices. Then, we combine the completed sub-matrices entries to produce the whole spectrum occupancy matrix. Through simulations, we show that the proposed framework efficiently achieves high detection in the sensing step and minimizes the spectrum occupancy matrix recovery error while reducing the overall network overhead.

我们提出了一种新颖的框架，用于实现可伸缩的数据库驱动的动态频谱访问以及异构宽带频谱的共享。拟议的框架由两种互补的方法组成，这些方法利用了压缩感测理论，低秩矩阵理论和用户合作的优势，通过克服占用频带数的时变性来建立准确的异构宽带频谱图。每个传感节点（SN）的测量数量众多，固有无线信道的损害以及较高的报告网络开销。首先，利用附近的SN具有高度相关的频谱观测这一事实，我们利用分布式压缩感测仅通过每个SN的少量测量即可实现协作异构宽带频谱感测。第二，为了减少由于所关注频谱的高宽度而导致的网络开销，我们提出了一种两步方法，该方法使用占用子矩阵的本地低秩属性来执行频谱占用恢复。然后，我们组合完成的子矩阵条目以产生整个频谱占用矩阵。通过仿真，我们表明，所提出的框架在感测步骤中有效地实现了高检测，并且在降低总体网络开销的同时，最大程度地降低了频谱占用矩阵的恢复误差。