标题: 利用振幅资讯之单一位元压缩式感测 应用于分散式参数估计技术
Amplitude-Aided 1-Bit Compressive Sensing Over Noisy Wireless Sensor Networks
作者: 陈卿衔
Chen, Ching-Hsien
吴卓谕
Wu, Jwo-Yuh
电信工程研究所
关键字: 压缩式感测;量化;无线感测网路;Compressive Sensing;Quantization;Wireless Sensor Networks
公开日期: 2015
摘要: 本硕士论文是压缩式感测(Compressive Sensing, CS)应用于无线感测网路 (Wireless Sensor Network, WSN)上的一个研究,让压缩式感测使用量化过的量测值来估计原始的稀疏讯号。
压缩式感测是一项很常应用于无线感测网路中的技术,藉以对抗其实际环境的严苛限制。然而,感测值在传送前一定要经过量化才可以以数位传输的方式传送,然而量化的步骤随着量化的复杂度越高而越耗能的。为了节省资源,发展至今,甚至使每个量测值经过量化器后只被配与一个位元来代表唯二的代表值。而使压缩式感测利用被二元量化后的量测值来估计原始稀疏讯号,称单一位元之压缩式感测(1-Bit CS)。
在本硕士论文中,我们探讨将单一位元之压缩式感测应用在有杂讯的无线感测网路上,并且各个传送端将其位元送到中央接收端时会有位元翻转的可能。于是我们提出了一套以振幅的资讯来辅助重建原讯号的方法,可分成两个部分:1. 设计量化代表点,使中央接收端得到的量测值,因为量测杂讯、量化误差以及位元翻转的均方误差能够达到最小。2. 中央接收端将得到的量测值经过专门重建稀疏向量的 最小化的重建方法 ( -Minimization Method)来估计原稀疏向量。在最后得到与量化代表点、原始讯号、量测杂讯以及位元翻转机率有关的均方误差式子,并且可以得到以闭合型式解的最佳均方误差量化代表点。
在模拟中可观察到,我们所提出的方法可在感测时的讯号杂讯比较低以及感测器数目较少的时候,比现行很多专门设计给单一位元之压缩式感测的叠代重建方法有更高的准确度。
One-bit compressive sensing (CS) is known to particularly suited for resource-constrained wireless sensor networks (WSNs). In this paper, we consider 1-bit CS over noisy WSNs subject to channel-induced bit flipping errors, and propose an amplitude-aided signal reconstruction scheme, by which (i) the representation points of local binary quantizers are designed to minimize the loss of data fidelity caused by local sensing noise, quantization, and bit sign flipping, and (ii) the FC adopts the conventional -minimization method for sparse signal recovery using the decoded and de-mapped binary data. The representation points of binary quantizers are designed by minimizing the mean square error (MSE) of the net data mismatch, taking into account the distributions of the nonzero signal entries, local sensing noise, quantization error, and bit flipping; a simple closed-form solution is then obtained. Numerical simulations show that our method improves the estimation accuracy when SNR is low or the number of sensors is small, as compared to state-of-the-art 1-bit CS algorithms relying solely on the sign message for signal recovery.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT070260235
http://hdl.handle.net/11536/126359
显示于类别:Thesis