中文摘要：

针对现有认知传感网在低信噪比条件下感知性能差、感知时间长、协作开销大的问题,提出了一种双重序贯检测的协作频谱感知（CSDS）方法.该方法首先在各传感器节点将数据进行分段处理,以各分段的能量作为一个样本点进行序贯检测,当各节点的对数似然比和大于门限时向融合中心发送协作信息,然后在融合中心再进行序贯检测直到作出判决.理论推导给出了平均样本数、平均感知时间等性能参数的精确表达式.采用Matlab的仿真结果表明：CSDS方法在低信噪比条件下,在保持感知性能的同时,所需平均样本数和平均感知时间低于能量检测的40％,且协作开销较现有协作序贯检测方法显著减少.

英文摘要：

To solve the problems of spectrum sensing with low signal-noise ratio （SNR） in cognitive radio sensor networks （CRSN）, a cooperative spectrum sensing method by dual sequential detection （CSDS） is proposed, where the data in each sensor are processed in segments. A sequential probability ratio test （SPRT） is taken in each sensor, and a sensor sends cooperative messages to fusion center （FC） when the sum of logarithm likelihood ratio gets beyond the threshold. Then another SPRT is taken in FC until to make a decision. Performance analysis of the average sample number （ASN） and average sensing time （AST） are also carried out. Simulations based on Matlab validate the effectiveness of the method in the low SNR regime. ASN and AST of CSDS present below 40% of those by energy detection, and the cooperative overhead greatly decreases compared with other cooperative sequential detection.