单通道混合信号粒子滤波盲分离算法的快速实现

High Speed Implementation for Particle Filtering Based Single-channel Blind Separation Algorithm

粒子滤波可以处理非线性非高斯问题,可以应用于混合信号的盲分离中,是一种有效的盲分离算法。但是粒子滤波算法存在一个很大的缺陷,其复杂度太大,为了减少粒子滤波运算时间、提高运算效率,通过对粒子滤波实现过程中的符号向量进行前置扩维,从而改进了粒子滤波实现方法。仿真结果表明:改进后的算法在运算速度上相比之前有了比较明显的提升。

Particle filtering can deal with nonlinear and non-Gaussian problem.It＇s an effective blind separation algorithm and can be applied to the blind separation of mixed-signal.But high computational complexity is a great flaw of particle filtering.The implementation method of particle filtering is analyzed and improved,based on preposing and dimension expansion of symbol vector,so that particle filtering algorithm＇s operation time can be reduced and computation efficiency can be improved.Simulation results show that compared with the traditional algorithm,the improved algorithm has a significant improvement in speed.