非高斯噪声下信号盲检测算法*

针对实际甚低频和超低频接收机不仅受非高斯噪声的影响,同时受到接收机内部和外部环境中高斯噪声影响的问题,对噪声采用高斯尺度混合分布和高斯分布的混合模型建模,根据混合模型的性质,设计了一种基于马尔可夫链蒙特卡罗方法的信号盲检测算法。盲检测算法在贝叶斯层次模型下,采用Gibbs抽样和M-H抽样更新参数,同步检测信道衰落系数、噪声模型参数和信号。算法迭代效率快、精度高。通过与最优检测性能比较,盲检测算法性能优异,对甚低频和超低频信号接收具有重要的现实意义。

A blind signal detection algorithm in Non-Gaussian noise

Non-Gaussian noise is the main interference in very low frequency(VLF) and super low frequency(SLF) communication system. Considering receiver is not only affected by the non-Gaussian noise but also Gaussian noise, a mixed model composed by Gaussian scale mixture(GSM) distribution plus Gaussian distribution is proposed. A blind detection algorithm based on Markov chain Monte Carlo(MCMC) algorithm is designed according to the properties of the mixed model. The blind detection algorithm can detect the channel fading coefficient, parameters of noise model, and signals at the same time through Gibbs sampler and M-H algorithm, which is based on Bayesian hierarchical model. The algorithm has a high Iterative efficiency and precision. The results show that the proposed blind detection algorithm performs as well as the optimal detection and can be excellently applied in practice.