Contributions to energy-aware demand-response systems using SDN and NFV for fog computing
Ever-increasing energy consumption, the depletion of non-renewable resources, the climate impact associated with energy generation, and finite energy-production capacity are important concerns worldwide that drive the urgent creation of new energy management and consumption schemes. In this regard,...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Estado: | Versión publicada |
| Fecha de publicación: | 2022 |
| País: | España |
| Institución: | CBUC, CESCA |
| Repositorio: | TDR. Tesis Doctorales en Red |
| OAI Identifier: | oai:www.tdx.cat:10803/674381 |
| Acceso en línea: | http://hdl.handle.net/10803/674381 https://dx.doi.org/10.5821/dissertation-2117-368245 |
| Access Level: | acceso abierto |
| Palabra clave: | Àrees temàtiques de la UPC::Enginyeria de la telecomunicació 004 621.3 |
| Sumario: | Ever-increasing energy consumption, the depletion of non-renewable resources, the climate impact associated with energy generation, and finite energy-production capacity are important concerns worldwide that drive the urgent creation of new energy management and consumption schemes. In this regard, by leveraging the massive connectivity provided by emerging communications such as the 5G systems, this thesis proposes a long-term sustainable Demand-Response solution for the adaptive and efficient management of available energy consumption for Internet of Things (IoT) infrastructures, in which energy utilization is optimized based on the available supply. In the proposed approach, energy management focuses on consumer devices (e.g., appliances such as a light bulb or a screen). In this regard, by proposing that each consumer device be part of an IoT infrastructure, it is feasible to control its respective consumption. The proposal includes an architecture that uses Network Functions Virtualization (NFV) and Software Defined Networking technologies as enablers to promote the primary use of energy from renewable sources. Associated with architecture, this thesis presents a novel consumption model conditioned on availability in which consumers are part of the management process. To efficiently use the energy from renewable and non-renewable sources, several management strategies are herein proposed, such as the prioritization of the energy supply, workload scheduling using time-shifting capabilities, and quality degradation to decrease- the power demanded by consumers if needed. The adaptive energy management solution is modeled as an Integer Linear Programming, and its complexity has been identified to be NP-Hard. To verify the improvements in energy utilization, an optimal algorithmic solution based on a brute force search has been implemented and evaluated. Because the hardness of the adaptive energy management problem and the non-polynomial growth of its optimal solution, which is limited to energy management for a small number of energy demands (e.g., 10 energy demands) and small values of management mechanisms, several faster suboptimal algorithmic strategies have been proposed and implemented. In this context, at the first stage, we implemented three heuristic strategies: a greedy strategy (GreedyTs), a genetic-algorithm-based solution (GATs), and a dynamic programming approach (DPTs). Then, we incorporated into both the optimal and heuristic strategies a prepartitioning method in which the total set of analyzed services is divided into subsets of smaller size and complexity that are solved iteratively. As a result of the adaptive energy management in this thesis, we present eight strategies, one timal and seven heuristic, that when deployed in communications infrastructures such as the NFV domain, seek the best possible scheduling of demands, which lead to efficient energy utilization. The performance of the algorithmic strategies has been validated through extensive simulations in several scenarios, demonstrating improvements in energy consumption and the processing of energy demands. Additionally, the simulation results revealed that the heuristic approaches produce high-quality solutions close to the optimal while executing among two and seven orders of magnitude faster and with applicability to scenarios with thousands and hundreds of thousands of energy demands. This thesis also explores possible application scenarios of both the proposed architecture for adaptive energy management and algorithmic strategies. In this regard, we present some examples, including adaptive energy management in-home systems and 5G networks slicing, energy-aware management solutions for unmanned aerial vehicles, also known as drones, and applicability for the efficient allocation of spectrum in flex-grid optical networks. Finally, this thesis presents open research problems and discusses other application scenarios and future work. |
|---|