基于群Monte Carlo的大气噪声二维模型参数估计

为解决多天线最佳接收下的多维非高斯噪声参数估计问题，提出了基于群蒙特卡洛的大气噪声二维模型参数估计方案，通过联合设计蒙特卡洛马尔科夫链和优化重要性重采样算法，实现噪声模型的全局最优参数估计。针对该算法高强度运算需求，在GPU平台上对核心运算作细粒度并行计算处理并优化设计，使运算速度大幅提升，以满足实时处理要求。仿真实验结果表明，该算法迭代收敛快，精度高，各参数估计相对误差普遍小于0.02，最大相对误差可控制在0.05以内，运算速度较传统计算有大幅度的提高，可充分满足低频通信系统中实时计算的要求。

Parameter estimation of 2-D model for atmospheric noise based on population Monte Carlo algorithm

In order to solve the problem that includes the parameter estimation of the multi-dimensional non-Gaussian noise model with multi-antenna optimum receiver,a method is proposed to estimate parameters of two-dimensional(2-D) atmospheric noise model based on population Monte Carlo(PMC). Both the Markov chain Monte Carlo algorithm and optimized sampling importance resampling method are used to achieve the global optimal parameter estimation of the multi-dimensional non-Gaussian noise model. Besides,the corresponding algorithm is designed.In consideration of the algorithm requirement for low computational complexity,core computation is designed for fine grain parallelization based on the graphics processing unit(GPU). It improves the algorithm efficiency greatly,and can satisfy the need for real-time processing. The simulation results show that the presented algorithm possesses the characteristics of high precision and fast convergent iteration. The relative error is generally smaller than 0.02,and the maximum relative error is smaller than 0.05. Compared with traditional computing method,the presented method can improve the computing efficiency greatly. And it can fully satisfy the real-time computation in low frequency communication systems.