-
5月8日学术报告(University of Tennessee, Knoxville: Wei Gao)
-
类别:网络整理
发布人:admin
浏览次数: 次
发布时间:2015-09-02 14:25
-
报告日期及时间:2015年5月8日周五 14:30
报告地点: 计算机学院4楼B404报告厅
报告人: Wei Gao
报告人国籍: 中国
报告人单位:University of Tennessee, Knoxville
报告人简介:Wei Gao is an Assistant Professor of the Department of Electrical Engineering and Computer Science at the University of Tennessee,
Knoxville. He received his PhD degree in Computer Science from the Pennsylvania State University and BE degree from the
University of Science and Technology of China. His research interests include mobile and distributed computing systems, wireless
networking, and smart grid. He has published more than 40 papers at top-tier academic conferences and journals including ACM MobiHoc, IEEE INFOCOM, ICNP, ICDCS, and so on. He is a member of IEEE and has been serving at the technical
program committee of several international conferences including IEEE INFOCOM, ICNP, and ICDCS.。
报告摘要:
Mobile cloud computing (MCC) bridges the gap between the limited capabilities of mobile devices and the increasing complexity
of mobile applications, by offloading the local computations to the cloud. In this talk, I will present our recent work on exploiting
the mobile system dynamics to address two fundamental problems in MCC: i) what to offload, and ii) how to offload. To address
the first problem, we develop analytical models to formulate and exploit the dynamic patterns of mobile applications’ run-time execution for workload offloading. To ensure the energy efficiency of MCC, traditional schemes partition a mobile application
and only offload the portion that benefits the most from MCC to the cloud. However, the portion being offloaded is decided only
through offline program analysis, and may be inappropriate at run-time due to the heterogeneity of applications’ execution. Our approach takes such run-time heterogeneity into account through a semi-Markov-based formulation, and makes offloading decisions accordingly based on a probabilistic framework. Furthermore, we address the second problem by systematically
enforcing the offloading decisions and minimizing the amount of memory contexts being migrated to the cloud. Our approach
builds on fine-grained program parsing, which tracks the minimum set of memory contexts that are relevant to remote execution of mobile program methods. The parsing results are then taken as input to support run-time migration of these program methods to the cloud. Our schemes have been integrated to the kernel of Android OS and operate directly over the mobile
application executables.
邀请人: 王骞教授 王志波副教授
-
- 上一篇:5月11日学术报告(伊利诺伊大学厄巴纳-香槟分校Prof. Peng Jian )
- 下一篇:2015年5月12日学术报告信息(Simon Fraser University:Jiangchuan Liu )