摘要
为提高弱相关性网络数据压缩感知的可靠性和有效性,提出一种基于迭代凸优化的网络数据重构方法。该方法利用多次凸优化算法共同重构相关性较弱的网络数据,在每次运行凸优化算法后,对已重构出的数据向量元素进行加权以降低权值,而使其他的数据向量元素在下次凸优化中得到重构。与以往的压缩感知重构方法相比,迭代凸优化重构可在网络数据相关性较弱的情况下保证重构准确度。仿真实验验证了所提方法的正确性。
For improving the reliability of Compressed Sensing （CS） reconstruction algorithms for Wireless Sensor Network （WSN） data under weak correlation, an iterative convex optimization based recovery algorithm is proposed in this paper. The iterative algorithm reconstructs the coefficients by several convex optimizations and the weights of the nonzero signal coefficients which is identified in early iterations is down-weighted in order to allow more sensitivity for identifying the remaining small nonzero signal coefficients. Compared with the existed algorithm for network data recovery, the algorithm can ensure the accuracy of recovery under weak correlation. The performance of the algorithm is validated by using simulations.
出处
《电视技术》
北大核心
2013年第9期99-102,122,共5页
Video Engineering
关键词
无线传感器网络
压缩感知
相关性
迭代凸优化
wireless sensor network
compressed sensing
correlation
iterative convex optimization
作者简介
常国锋（1978-），讲师，硕士，主研无线传感器网络；
张军（1982-），讲师，硕士，主研计算机应用等。
