基于非参量CUSUM的认知超宽带频谱感知算法

提出了一种基于非参量CUSUM的超宽带信号快速检测算法，该算法针对认知超宽带频谱感知中超宽带信号检测问题，根据超宽带信号占空比低的特点，选择短时能量作为非参量CUSUM算法的观测统计量，将块检测思想和序列检测思想进行有机结合，克服了块检测算法的信噪比门限效应，并采用加滑动预处理窗的方式降低噪声对观测统计量的影响。本文通过自适应改变非参量CUSUM算法中Reference Value取值的方法，提出改进算法以进一步缩短检测延迟。仿真证明，占空比为1/10、信噪比为-25dB时, 所提算法可在用户发送50个脉冲的时间内以90%的概率检测出主用户信号的存在，而能量检测算法在8dB时才能达到相同检测概率，且所提算法检测延迟小于能量检测算法。改进算法在-10dB≤SNR≤5dB条件下可明显缩短检测延迟。 

Cognitive UWB Spectrum Sensing Algorithm Based On Nonparametric CUSUM

Based on nonparametric CUSUM algorithm, an UWB signal quickest detection algorithm is proposed, to solve the problem of UWB signal detection in CUWB (cognitive UWB) system. According to the low duty ratio property of UWB signal, short time energy is selected as the observation statistic to connect the block detection idea and the sequence detection idea, which conquers the SNR wall problem. A slipped pretreatment window is adopted to depress the noise interference to observation statistic. To more decrease the detection delay, the Reference Value of CUSUM algorithm is adaptively changed in improved algorithm. It is improved by simulation that the proposed algorithm can achieve more than 90% detection probability, under the condition that duty ratio is 1/6 and SNR= -25dB, within 50 transmitted pulses, and energy detection algorithm can achieve the same detection probability when SNR=8dB. Under the condition SNR -10dB, the improved algorithm can obviously decrease the detection delay.