Revisiting Model Order Selection: A Sub-Nyquist Sampling Blind Spectrum Sensing Scheme
Hui Ma,Xiaobing Yuan,Jiang Wang,Baoqing Li
DOI: https://doi.org/10.1109/twc.2022.3217764
IF: 10.4
2023-05-12
IEEE Transactions on Wireless Communications
Abstract:Wideband spectrum sensing based on sub-Nyquist sampling is an attractive approach to advance dynamic spectrum sharing (DSS), which can improve frequency resource utilization while overcoming sampling bottlenecks. Under the Compressive Sensing (CS) framework, finding occupied subbands can be equivalent to computing the support set for the Multiple Measurement Vectors (MMV) problem. To guarantee the performance of joint support recovery in noisy environments, different kinds of prior information are required, one of which is sparsity, a time-varying parameter. To address the dependence of recovery performance on signal sparsity, this paper proposed a two-step scheme for blind wideband spectrum sensing using a Modulated Wideband Converter (MWC) sub-Nyquist sampling front-end. The scheme first adopts the model order selection (MOS) method to estimate sparsity from the compressed covariance matrix, and then uses the estimates to dynamically adjust joint support recovery. The complete theoretical derivation innovatively applies MOS to sub-Nyquist sampling and presents a design method for MOS penalty constant. Extensive simulation results show that the proposed scheme can not only achieve blind sensing under the spectrum occupancy up to 40%, reduce the overall computational complexity of iterative SOMP, but also significantly improve the false alarm performance, meeting the requirements of the IEEE 802.22 standard.
telecommunications,engineering, electrical & electronic