FESB: Improving system energy efficiency by leveraging context awareness
Pomoć za Pretraži
 
 
Slika Lelas Damir
Improving system energy efficiency by leveraging context awareness
napisao/la Lelas Damir - Subota, 12 Travanj 2014, 13:09
 

Zavod za elektroniku i Kompetencijski centar za programsko inženjerstvo pozivaju vas na predavanje u utorak 15. travnja, u 12:15 u dvorani A100 koje će održati prof. Tajana Šimunić Rosing.

Više detalja možete pronaći u nastavku.

Abstract:

The proliferation of personal computing and the advent of cheap, small sensors have given the rise of computing at the edge of the traditional computational infrastructure. While various technological components of the computing systems at the edge of internet already exist, the key to success of this new class of systems are advances in the abstractions that can support easy extensibility, and development of adaptive energy management strategies that ensure efficient system operation. In this talk I first give an overview of the systems and algorithms we have developed at UCSD to enable the development of adaptive edge computing infrastructure along with strategies that significantly lower the energy consumption in sensing, mobile and edge server infrastructures. The rest of the talk focuses on how context can be leveraged to enhance the system operation. Context is defined as any relevant information that provides a value-add to improving energy efficiency. Source-side context, in the form of environmental variables, can help predict solar and wind output, while load-side context, such as workload analysis and prediction, can tailor the load to accommodate the energy source variability. We show how we can leverage context in two applications: data centers and individual residences with on-site renewables and distributed energy storage. By applying context data to automation, prediction, and smart scheduling, we demonstrate over 90% improvement in green energy efficiency for data centers, and over 40% improvement in green energy costs in residences when compared with using no context.

Bio:
Tajana Šimunić Rosing is currently a Professor of Computer Science and Adjunct Professor in the Electrical and Computer Engineering Department at UCSD. She is currently heading the effort in SmartCities as a part of DARPA and industry funded TerraSwarm center. Prior to that she led the energy efficient datacenters theme as a part of the MuSyC center. Her research interests are energy efficient computing, embedded and large scale distributed systems. Prof. Rosing was a full time researcher at HP Labs while being leading research part-time at Stanford University. She finished her PhD in 2001 at Stanford University, concurrently with finishing her Masters in Engineering Management. Her PhD topic was Dynamic Management of Power Consumption. Prior to pursuing the PhD, she worked as a Senior Design Engineer at Altera Corporation.