摘要
该文在并行次梯度投影技术(Parallel Subgradient Projection,PSP)的基础上,提出一种加速回波抵消算法。加速算法通过一种"角度寻优"策略,对由分属不同状态下的输入信号组成的空间进行"寻优",从而找到一对最优投影空间。然后向这一对最优半空间的交集投影,实现对自适应算法的加速。实验仿真表明,加速算法相比传统方法收敛速率更快,无论是用均方误差还是自适应滤波器失准(Echo Path Misalignment)指标衡量加速算法在性能上都有一定程度的提升。
An accelerated acoustic echo cancellation algorithm is proposed in this paper based on parallel Subgradient Projection (PSP) technique. Through the “angle optimization” strategy, accelerated method optimizes the input signal spaces belong to different states and finds a pair of optimal projection spaces. Then adaptive algorithm can be accelerated by projection onto the pair of optimal half spaces. Experiment indicates the accelerated method possesses more rapid convergence speed than traditional algorithm. Additional, system performance gets promoted no matter evaluated by mean square error or echo path misalignment.
出处
《电子与信息学报》
EI
CSCD
北大核心
2008年第5期1140-1143,共4页
Journal of Electronics & Information Technology
基金
国家自然科学基金项目(60472058)
教育部博士点基金(20050286001)
教育部新世纪优秀人才支持计划(NCET-04-0483)资助课题
关键词
声回波抵消
并行次梯度投影
自适应滤波
Acoustic echo cancellation
Parallel Subgradient Projection (PSP)
Adaptive filtering