基于盖氏圆方法的宽带信源数目估计

Detection of number of wideband sources based on gerschgorin disks

针对低信噪比和小快拍数情况下宽带信号源数目较难精确估计的问题,提出了一种基于盖氏圆方法的宽带信号源数目估计算法(GDECSM)。盖氏圆方法通过利用盖氏半径对矩阵特征值范围的限定关系,能有效区分信号子空间与噪声子空间,在宽带信号源数目估计中能够产生良好的检测效果。仿真实验表明,与基于AIC准则的宽带信源数目估计算法(AICCSM)相比,GDECSM算法在低信噪比和小快拍数时都具有更高的检测概率,运算复杂度也有所降低。

To solve the problem of exact estimation of wideband sources at low signal noise ratio（SNR） or little snapshot, a new method to determine the number of wideband coherent signals based on Gerschgorin disks was proposed. Gerschgorin disks method could effectively partition the signal subspace and noise subspace using the constraining relation of the Gerschgorin radii to eigenvalue of the matrix, and as a result, a good detection performance could be produced. Simulation results show that the new method improves the probability of detection at low SNR and little snapshot and reduces the computational load compared to the method to determine the number of wideband coherent signals based on Akaike information criterion（AIC）.