基于多尺度功率谱子带梯度的宽带频谱感知算法与性能分析

为了应对复杂环境下非合作通信、电磁频谱监管等宽带接收中存在的先验信息缺失、信道失真严重以及频域呈现不平坦色噪声的挑战，提出一种基于多尺度功率谱子带梯度的宽带频谱感知算法，该算法不要任何的先验信息，对功率谱进行分段计算梯度，再进行自适应双阈值检测，通过多尺度的技巧提高了宽带频谱感知的稳定性。对该算法在不同信道模型下的统计特性、虚警概率、检测概率以及判决门限的表达式进行了理论推导。理论分析和实验仿真表明，算法适用于高斯噪声信道和平坦衰落信道，能够有效克服色噪声，并且能够实现用户频带范围定位，运算复杂度低、实时性强，对噪声不确定度具有稳健性，能够用于低信噪比场合。

Multi scale power spectral density subband gradient-based spectrum sensing algorithm and performance analysis

In response to the challenge of the complex environment such as non-cooperative communication and wideband electromagnetic spectrum regulation, where existing missing priori information, serious channel distortion and uneven color noise in frequency domain, a novel spectrum sensing algorithm based on multi-scale power spectral density subband gradient(MPSG) was proposed. The proposed algorithm, not relying on any prior information, calculated gradients of sections for the power spectrum and detected the narrowband signals by adaptive double threshold, which improved the stability of wideband spectrum sensing through multi-scale technique. Theoretically expressions of statistical properties, false alarm probability, detection probability and decision threshold in different channels were deduced. The theoretical analysis and simulation results prove that the novel algorithm overcomes effectively the color noise with low computational complexity and strong real-time in Gaussian noise and fading channel, which can accomplish occupying band range localization of prime users. In addition, the proposed algorithm is robust for noise variance uncertainty, even in the low SNR environment.