摘要
基于融合中心的多观测向量联合稀疏信号恢复算法需要将各个传感节点的数据传输到融合中心(融合中心可能远离各个节点),该方法在节点功率受限以及缺少融合中心的传感网络中并不适用。为了克服上述困难,本文提出了一种分布式路径协同优化算法来解决上述问题。由于采用了分布式计算和路径协同优化,各个传感节点只需与其近邻节点进行少量的数据交互,每个节点所消耗的传输数据功率和所承受的计算复杂度较低。实验结果表明,本文提出的算法的性能能够很好的逼近基于融合中心的联合稀疏信号恢复算法的性能。
Joint-sparse recovery from multiple measurement vectors has to transfer all the measurement vectors obtained from different nodes or sensors to fusion center ( maybe far away from individual nodes). However, collecting all the data to fusion center (FC) may be challenging or impossible, especially in the cases that the power and computing resources are limited, or there is no FC. To overcome the above problem, a distributed pathwise coordinate optimization algorithm is de-veloped to solve joint-sparse from multiple measurement vectors (MMV). In the proposed scheme, MMV problem is refor-mulated into separable forms, which can be solved distributed. To further reduce the complexity of the algorithm, pathwise coordinate optimization (PCO) algorithm is used to approximately solve the separable forms. Benefiting from the distributed computation and PCO, the new algorithm entails low computation and power overhead, and affordable data transferring for each node among its neighbors. Simulation results show that the new distributed algorithm is very competitive to the central-ized algorithm on the performances of sparsity recovery and support detection.
出处
《信号处理》
CSCD
北大核心
2013年第8期964-970,共7页
Journal of Signal Processing
基金
国家自然科学基金(60872073
6097501
51075068)
教育部博士点专项基金(20110092130004)资助课题
关键词
压缩感知
联合稀疏信号恢复
多观测向量
路径协同优化
分布式计算
Compressed Sensing
Joint-Sparse Recovery
Multiple Measurement Vectors
Pathwise Coordinate Optimiza-tion
Distributed Algorithm