基于奇异值分解的宽带频谱感知改进算法

对认知无线电中宽带频谱感知的信号检测问题进行研究，在信道位置和带宽随机分布的条件下，提出一种基于Hankel矩阵奇异值分解的改进算法完成宽带频谱的频率点奇异性检测并获得各子频段的起始频率和带宽，有效实现盲宽带频谱感知，同时利用提出的三种性能评价参数对该算法与多尺度小波变换算法进行了性能对比。通过实际接收微波信号及仿真OFDM信号感知实验验证，该算法有效抑制了噪声不确定性，滤除了伪奇异点，提高了宽带频谱感知性能。

Improved algorithm based on Singular Value Decomposition in wideband spectrum sensing

The signal detection issue is investigated for wideband spectrum sensing in cognitive radio. Considering that channel position and bandwidth are randomly distributed, an improved singularity detection algorithm based on second-order component of hankel matrix Singular Value Decomposition is proposed for real-time detection of multiple signals. Applied to frequency points singularity detection in the entire frequency range, the start-end frequencies and the bandwidths of each band can be obtained. On this basis, traditional multi-band detection in fixed channel is used to classify bands as occupied or vacant, so that it can efficiently accomplish wideband sensing, meanwhile. With the three new evaluated parameters, experimental results based on actual received microwave signal and emulational OFDM signal sustain that the proposed algorithm is better than multiresoluted wavelet in singularity detection.