الفهرس 
Regularized Modified Basis Pursuit DenoisingOnly 14 pages are availabe for public view
 |  | from | 144 |  |  |
| | | from | 144 | | |
Abstract
Compressive sensing (CS) was proposed as a promising approach to reduce data acquisition cost by enabling sub-Nyqusit sampling. CS enables reconstruction of the signal with fewer number of measurements provided that the signal has a sparse representation in some domain. CS has been applied in many wireless applications such as wideband spectrum sensing in cognitive radio networks (CRNs), data gathering in internet of things (IoT) and wireless sensor networks (WSNs), and channel estimation in massive MIMO systems. In many of these applications, incorporating side information with CS can lower the bounds on the number of measurements needed for successful reconstruction of the sparse signal by adding a prior in the reconstruction process. In most of applications, side information can be acquired such as a prior knowledge of the signal values or a known signal structure.
In this thesis, we investigate the utilization of compressive sensing with side information in wideband spectrum sensing and IoT/WSN compressive data gathering. In addition to the sparse structure of the signal, our approaches in compressive spectrum sensing exploit the temporal structure of the spectrum. In compressive data gathering we utilize the information from existing operation models to enhance the compressive sensing performance. We can summarize the work presented in this thesis into two main parts.
The first part of this thesis approaches the problem of compressive spectrum sensing in cognitive radio networks. It exploits the temporal structure arising from the regularity of primary user (PU) traffic patterns. This contrast to most of the work in compressive spectrum sensing that solely exploits the spatial and frequency domain structure of the spectrum neglecting the temporal domain. The effectiveness of incorporating PU traffic patterns in compressive spectrum sensing is investigated. This achieves improved sensing performance through exploiting the statistics of the PU activity in the CS recovery algorithms. Experimental analysis through simulation shows that the proposed schemes can substantially improve the receiver operating characteristic (ROC) performance even at noisy spectrum measurements taken at lower sampling rate.
The second part of this thesis studies compressive data gathering in large scale dense IoT networks. Smart city applications are becoming among the fastest growing segments of government. One of their highly promising applications is the continuous dense monitoring of critical infrastructure such as water, gas, and electrical networks. Technology advancements have made it possible to create low cost battery-operated Internet of things (IoT) devices to monitor the operating parameters of these networks. However, the challenge remains to increase the lifetime of these sensors and reduce its communication burden on the network. In this thesis, a compressive data gathering approach is proposed that utilizes information from existing operation models of these networks and combines it with the temporal and spatial correlation of the measured data in a compressive sensing (CS) framework. The proposed approach can effectively reduce the required number of measurements and achieves a more efficient node activation strategy to extend the overall network lifetime. We give an example of the effectiveness of our approach using a hydraulic model for water distribution networks. We evaluate the results of the proposed compressive data gathering approach using the Hanoi water network data set and a pilot area in Hurghada, Egypt.