자료유형 | 학위논문 |
---|---|
서명/저자사항 | From 2.5G to 5G: Enhancing Access and Performance for Mobile Users. |
개인저자 | Ahmad, Talal. |
단체저자명 | New York University. Computer Science. |
발행사항 | [S.l.]: New York University., 2019. |
발행사항 | Ann Arbor: ProQuest Dissertations & Theses, 2019. |
형태사항 | 122 p. |
기본자료 저록 | Dissertations Abstracts International 81-06B. Dissertation Abstract International |
ISBN | 9781392850039 |
학위논문주기 | Thesis (Ph.D.)--New York University, 2019. |
일반주기 |
Source: Dissertations Abstracts International, Volume: 81-06, Section: B.
Advisor: Subramanian, Lakshminarayanan. |
이용제한사항 | This item must not be sold to any third party vendors. |
요약 | This dissertation has two overarching themes: i) enhancing access to connectivity for mobile users in rural contexts and ii) enhancing transport layer performance for mobile users.More than half of the world's population faces barriers in accessing the Internet. A recent ITU study estimates that 2.6 billion people cannot afford connectivity and that 3.8 billion do not have access to it. To enhance access I have worked on two projects: Wi-Fly and GreenApps. Wi-Fly is a new connectivity paradigm designed for regions without Internet coverage that enables communication between lightweight Wi-Fi devices on commercial planes and ground stations. Through empirical experiments with test flights and simulations, we show that Wi-Fly and its extensions have the potential to provide connectivity in the most remote regions of the world. In GreenApps, we look at how localized cellular applications can be built for rural communities on top of software-defined cellular base stations. We deployed the GreenApps platform on rural base stations for communities in Ghana and Nicaragua, and supported multiple localized applications for rural communities.Enhancing transport layer performance over cellular networks is critical to improve end-to-end application performance for mobile users. Cellular networks have unique challenges that make conventional transport protocols unsuitable for these environments. In the past few years, several new delay-based congestion control algorithms have been developed with complex nonlinear control loops for cellular contexts. While these protocols have shown promise, it has been extremely challenging to analyze and interpret the behavior of these algorithms especially under highly variable network conditions. In the Model-Driven Interpretable (MDI) congestion control work, we provide a model-driven framework to reason about the behavior of such congestion control algorithms. Our modeling approach simplifies a congestion control algorithm's behavior into a guided random walk over a two-dimensional Markovian state space(a Markov model). We show that the model of a congestion control algorithm can give key insights into its convergence and performance. More recently, we also looked at how to learn early signals of congestion in highly varying 5G channels. In particular we worked with Wi-Gig traces collected at 60 GHz and showed that it is possible to learn highly accurate early congestion signals using delay features observed at end hosts. |
일반주제명 | Computer science. |
언어 | 영어 |
바로가기 | ![]() |