CAJ | 학술논문

针对认知无线电网络中单节点的宽带压缩频谱感知算法检测准确性低、实时性差的缺点，提出了基于多测量向量块稀疏贝叶斯学习-快速边缘似然函数最大化(MBSBL-FMLM)的宽带协作频谱感知算法。该算法采用分布式压缩感知(DCS)系统进行多节点协作检测，以降低单节点检测带来的多径衰落、阴影衰落等不利影响；另外，融合中心结合多测量向量(MMV)模型和宽带信号的块稀疏结构得出多测量向量块稀疏贝叶斯学习(MBSBL)框架，并利用快速边缘似然函数最大化(FMLM)方法进行快速参数估计。数值分析表明，基于MBSBL-FMLM算法的检测概率、归一化均方误差、检测时耗均优于SOMP算法。
침대인지무선전망락중단절점적관대압축빈보감지산법검측준학성저、실시성차적결점，제출료기우다측량향량괴희소패협사학습-쾌속변연사연함수최대화(MBSBL-FMLM)적관대협작빈보감지산법。해산법채용분포식압축감지(DCS)계통진행다절점협작검측，이강저단절점검측대래적다경쇠락、음영쇠락등불리영향；령외，융합중심결합다측량향량(MMV)모형화관대신호적괴희소결구득출다측량향량괴희소패협사학습(MBSBL)광가，병이용쾌속변연사연함수최대화(FMLM)방법진행쾌속삼수고계。수치분석표명，기우MBSBL-FMLM산법적검측개솔、귀일화균방오차、검측시모균우우SOMP산법。
To improve detection accuracy and meet real-time detection of single node wideband compressed spectrum sensing algorithm in cognitive radio,a wideband cooperative spectrum sensing algorithm is presented.The proposed algorithm is based on multiple measurement vectors (MMV)block sparse Bayesian learning-fast marginalized likelihood maximization (MBSBL-FMLM),which uses multi-node cooperative detection method to reduce multipath fading and shadow fading.Mo-reover,fusion center combines MMV model and the block sparse structure of the wideband signal to obtain the MMV block sparse Bayesian learning (MBSBL),and utilizes FMLM method to estimate parameters rapidly.Numerical analysis indicates that the detection probability,NMSE and sensing time of MBSBL-FMLM algorithm are better than SOMP algorithm.