一种噪声未知的新型空间频谱分布协作感知算法

Novel cooperative sensing algorithm for spatial spectrum distribution with unknown noises

为了获得占用频段、主用户发射功率、位置和本地噪声等全局信息,给出了一种主用户全局功率谱模型,提出了一种噪声未知的新型空间频谱分布协作感知算法.利用变分贝叶斯推断理论,估计出模型系数向量和本地噪声向量,以求得全局信息.仿真结果表明,所提算法在较高信噪比下具有较高的估计精度和收敛稳定性.该算法性能虽弱于噪声已知算法的性能,但相比基于非负最小二乘准则的算法具有更好的均方误差性能.

In order to obtain the global information including occupied frequency bands,transmitting powers of the primary users（ PUs）,locations and local noises,a model for the global power spectral density（ PSD） of the PUs is constructed,and a novel cooperative sensing algorithm for spatial spectrum distribution with unknown noises is also proposed. By utilizing variational Bayesian inference（ VBI） theory,the model coefficient vector and the local noise vector are estimated to obtain the global information. The simulation results showthat the proposed algorithm has high accuracy and convergence stability with the high signal noise ratio（ SNR）. Though the performance of this algorithm is worse than that of the algorithm with known noises,but its mean square error（ MSE） performance is better than that of the algorithm based on the non-negativity least square（ NNLS） criterion.