Contributions to the solution of the energy efficient file distribution problem

  1. Verma, Kshitiz
Dirigida por:
  1. Alberto García Martínez Director

Universidad de defensa: Universidad Carlos III de Madrid

Fecha de defensa: 09 de noviembre de 2016

Tribunal:
  1. Fidel Cacheda Presidente/a
  2. Carmen Guerrero López Secretario/a
  3. Guillermo Ibáñez Fernández Vocal

Tipo: Tesis

Resumen

Concerns over the growing energy demands of humankind are increasing and energy efficiency is reported to be one of the major technological challenges of our times. The environmental problems related to Green House Gases have shown significant increase recently and the measures to counter its impact is reflected in policies of the governments and organizations. Apart from environmental effects, increasing energy demand has negative impacts on economy too. Owing to its importance, European Commission has been funding the energy efficiency projects very extensively and plans to invest more than 27 billion Euro to support the shift towards a low-carbon economy through European Structural & Investment Funds (ESIF). The Information and Communications Technology (ICT) sector is no exception. It has been shown through research that ICT is consuming energy comparable to the aviation sector and is still increasing rapidly. In order to address this issue, many energy efficient approaches applicable to ICT sector have been proposed in the literature. In this Thesis, we pick one of the most ubiquitous task in ICT, file distribution and concentrate on finding ways of transferring a file from one server to many hosts in the most energy efficient manner. We study the problem for one server and many host problem but our methods can be applied to many general scenarios including P2P file distribution, replication of content in a cloud, synchronization of caches in content distribution networks, downloading software updates to millions of PCs worldwide, and many more applications where the data disseminated does not have to be consumed instantaneously, for example, in video streaming. We show that our methods of file distribution are realistic and can save energy ranging from 50% to up to 10000% than the existing approaches. As compared to traditional methods, our proposed algorithms require strict collaboration between hosts but it is worth given the potential of energy savings. Given the ubiquity of file distribution, this can help reduce the energy consumption of ICT sector significantly.