透過您的圖書館登入
IP:3.149.242.118
  • 學位論文

使用遠程壓縮感測技術在有雜訊的物聯網

Remote Compressive Sensing for Noisy Machine-to-Machine Networks

指導教授 : 蘇炫榮

摘要


在最近幾年,互聯網 (M2M) 被廣泛應用於無限通訊系統中。機器本身會有功率限制, 處理和通訊能力也都會有限。壓縮感測技術可以避免傳多餘的資訊出去,進而壓縮傳送資料 量。在此篇論文中,在雙層架構下,我們於互聯網中針對隨機訊號源提出一個遠程的壓縮感 知架構,目的是壓縮閘道 (gateway) 的傳輸資料,並且我們將原本問題簡化成隨機壓縮感知 的問題。進一步,針對原本的系統中,在壓縮閘道到基地台中間,考慮通過一個白色雜訊的 通道的情況,讓系統更貼近真實生活的環境。 最後,我們找到兩種產生感知矩陣的方法。一個是奇異值分解共變異數矩陣 ( SVD Covariance Matrix),這方法是使用機器跟機器之間在空間上存在的關聯性來壓縮傳送資料。 另一個是適應性統計壓縮感知(Adaptive Statistical Compressive Sensing),這方法是參考過去傳 送過的資料,結合互信息來產生新的感知矩陣。將兩種方法應用於我們所提出的系統中,並 且驗證我們所選擇的這兩種感知矩陣在不同的系統下,可以達到傳統訊號傳輸所能達到的效 果優於之前的高斯 (Gaussian)或伯努力(Bernoulli) 感知矩陣。

並列摘要


In recent years, machine-to-machine (M2M) networks are widely con- sidered in wireless communication systems. To avoid the transmission of redundant information to improve the data rate, compressive sensing is a promising tool to be considered. In this paper under the two-tier architec- ture, we propose a remote compressive sensing scheme for the M2M networks with stochastic sources to improve the data rate and formulate a statistical compressive sensing problem. First we propose to use the minimum mean square error estimator at the gateway and the base station to transform the problem as a noisy statistical CS. We derive the form of a optimal decoder by following MMSE estimation for the proposed scheme. Furthermore We find two ways, that can produce the sensing matrices. There are SVD covariance matrix(SCM) and adaptive statistical compressive sensing(ASCS). SCM is using machines covariance matrix to compressed the data by reducing the correlation between each machines. ASCS uses the previous measurements and the sensing matrices obtained in the past states and combines and com- bine the mutual information to become a new method of CS.

參考文獻


[1] C.-H. Chang, H.-Y. Hsieh and H.-J. Su, “Not Every Bit Counts: Shifting
the Focus from Machine to Data for Machine-to-Machine Communica-
tions,” Asilomar Conference on Signals, Systems and Computers, Nov.
Cognitive Radiobased M2M Communications for Smart Meters,” Proc.
Y. Nozaki, “Toward Intelligent Machine-to-Machine Communications in

延伸閱讀