Most of today’s Internet of Things (IoT) applications assume that data will be moved offdevices into centralized cloud platforms. While existing IoT systems leverage cloud-based analytics for meaningful data reasoning, the assumption that data should always be moved off the devices is problematic. The amount of data to be moved from devices over Internet gateways to cloud platforms is huge which potentially make it cost inefficient. In other scenarios, privacy concerns of customers or organizational rules complicate the process of transferring data to third-party data centers.This dissertation proposes architectures and dynamic overlay network algorithms for in-networkand edge processing of data offered by the globally available IoT devices and provides a global platform for meaningful and responsive data analysis and decision making. The proposed techniques shift IoT analytics from a ”collect data now and analyze it later” scenario to directlyproviding meaningful information from the in-network processing of devices data at or near thedevices. The techniques serve future IoT use cases including distributed context awareness, on-demand data analysis, and in-network decision making. The dissertation comprises three main components.The first component is a device management protocol for cloning devices’ data in proximateEdge Computing platforms. Unlike existing application-layer IoT management protocols theproposed protocol uses the LTE LTE-A radio frame structure, device-to-device communication,and IoT data properties to avoid excessive network access latency in existing technologies.The second component realizes distributed IoT analytics as overlay networks of devices clones. By means of virtual network embedding, it selects and interconnects devices’ clones to efficiently realize applications’ virtual topologies to achieve goals such as minimum latency, minimum infrastructure cost, or maximum infrastructure utilization.Finally, the dissertation presents a communication middleware that allows autonomous discovery, self-deployment, and online migration of devices’ clones across heterogeneous Edge computing platforms. The middleware ensures that communication latency between clones is kept minimum despite the uncontrolled variability of the network and hosting platforms conditions.We evaluate the proposed architectures and algorithms through simulations and prototypeimplementation of various components in controlled testbed environments, which we evaluateusing real user applications. We explore the feasibility of the proposed techniques from boththeoretical and practical perspectives.
description.provenance : Approved for entry into archive by Julie Kurtz(email@example.com) on 2017-08-27T19:10:52Z (GMT) No. of bitstreams: 1AbdelwahabSherif2017.pdf: 17044219 bytes, checksum: 6680b1a41cd146be51f43c860b06f5e3 (MD5)
description.provenance : Made available in DSpace on 2017-09-06T18:32:14Z (GMT). No. of bitstreams: 1AbdelwahabSherif2017.pdf: 17044219 bytes, checksum: 6680b1a41cd146be51f43c860b06f5e3 (MD5) Previous issue date: 2017-08-04
description.provenance : Approved for entry into archive by Steven Van Tuyl(firstname.lastname@example.org) on 2017-09-06T18:32:14Z (GMT) No. of bitstreams: 1AbdelwahabSherif2017.pdf: 17044219 bytes, checksum: 6680b1a41cd146be51f43c860b06f5e3 (MD5)
description.provenance : Submitted by Sherif Abdelwahab (email@example.com) on 2017-08-26T12:21:12ZNo. of bitstreams: 1AbdelwahabSherif2017.pdf: 17044219 bytes, checksum: 6680b1a41cd146be51f43c860b06f5e3 (MD5)