采用分组数据的序贯频谱感知方法
A Sequential Spectrum Sensing Method Using Grouped Data

针对传统的认知无线电频谱感知方法在低信噪比下感知时间长、系统吞吐量低的问题,提出了一种采用分组数据的混合型序贯检测(MSD)方法。该方法首先将次用户的感知数据进行分组形成超采样,然后由数学理论推导出最大化系统吞吐量的最优虚警概率,并且利用牛顿迭代法搜索最优虚警概率,最后在最优虚警概率下对超采样序列依次进行细检测和粗检测,快速获得检测结果。MSD方法采用分组数据进行频谱感知,能有效缩短感知时长,获得最大系统吞吐量,从而提高频谱利用率。蒙特卡罗仿真结果表明,在低信噪比下MSD方法比传统的序贯检测法和序贯能量检测法的平均归一化吞吐量增加了109％和21％,平均感知开销率减少了75％和49％。
A mixed sequential detection (MSD) method using grouped data is proposed to solve the problems of long sensing time and low system throughput of conventional spectrum sensing methods under low signal？？noise ratio (SNR) condition. At first, the sensing data of the second user are processed in segments and grouped in super samples. Then, the optimal false alarm probability of the maximum system throughput is derived through mathematical theory analysis, and the Newton iterative algorithm is applied to search the optimal false alarm probability. Finally, the fine and rough detections on a sequence of super samples are successively taken under the optimal false alarm probability to quickly obtain the results of detection. The MSD performs spectrum sensing using grouped data, can effectively reduce the sensing time, and achieve the maximum throughput and improve the spectral efficiency. Monte Carlo simulation results under low SNR and comparisons with the sequential detection and the sequential energy detection show that the MSD gets 109%, 21% increase in average normalized throughput and 75%, 49% decrease in the ratio of average sensing overhead, respectively