Invited Speakers
Hong Shen
Efficient and Reliable Delivery of Cloud Services
Abstract
The emergance of various types of clouds makes efficient and reliable delivery of cloud services (content/storage/computation) increasingly important for effective access to cloud technologies. Service distribution between datacenters (facilities) of a cloud and clients can be mathematically formulated as the well-known facility location problem to compute a minimum-cost connection scheme for given client-facility requirements which is NP-hard even for the simplest case of uncapacitated connection. We show that efficient and reliable delivery of cloud services can be achieved by applying our solution to the problem of Fault-Tolerant Facility Allocation (FTFA) which extends fault-tolerant facility location (FTFL) by allowing each site (datacenter) to contain multiple replica facilities (stores/VMs) and incorporates the techniques of parallel access to the most economic site(s). Our algorithm applies inverse dual-fitting LP technique, runs in polynomial time and is provably within an approximation ratio of 1.52 to the optimal solution, matching the current best ratio for the classical uncapcitated facility location problem, a simpler version of FTFA. To achieve further improvement on efficiency for future accesses to the same service, we show that an effective scheme of en-route caching of service should be incorporated during delivery of the service from a datacenter to a set of clients it connects. Finally, we illustrate some other applications of FTFA, including service discovery in federated clouds, design of complex networks and fuzzy clustering.
Biography
Hong Shen is Professor (Chair of Computer Science) in University of Adelaide, Australia. Previously, he was Professor and Chair of Computer Networks Laboratory in Japan Advanced Institute of Science and Technology (JAIST) during 2001-2006, and Professor of Computer Science in Griffith University, Australia, where he taught 9 years. He has held numerous awards, titles, and special appointments, including "100 Talents" of Chinese Academy of Sciences, specially-appointed Professor in University of Science and Technology of China and in Beijing Jiaotong University. His main research interests lie in parallel and distributed computing, algorithms, high performance networks, privacy and security, data mining and multimedia systems. Received his Bachelor of Science from Beijing University of Science and Technology, Master of Science from University of Science and Technology of China, Ph.Lic. and Ph.D. from Abo Akademi University in Finland, Prof. Shen has published 250+ papers, with 100+ papers in international journals including a variety of IEEE and ACM transactions. He has served on editorial boards of 10 international journals, and chaired several international conferences.
Yuanchun Shi
What makes the Smart Space smart?
Abstract
Smart Space embodies pervasive computing in the scale of a room or a building. As a working or living environment with embedded computers, information appliances, and multi-modal sensors, allowing people to perform tasks efficiently by offering unprecedented information access and computing assistance, it has been regarded as typical system with the characteristics of pervasive computing. What core technologies make the Smart Space smart? In this talk, Prof. Shi will present the natural human computer interaction, device interoperation and context aware software infrastructure technologies for Smart Spaces. Also, some considerations will be drawn by analyzing the research aspects of pervasive computing.
Biography
Yuanchun SHI is a professor of the Department of Computer Science, the director of Institute of Human Computer Interaction & Media Integration of Tsinghua University, and the director of Pervasive Computing Division of Tsinghua National Lab of Information Science and Technology, China. She received her PhD, MS and BS in Computer Science from Tsinghua University. She was a senior visiting scholar at MIT AI lab during 2001-2002. She is an IEEE senior member. Her research interests include pervasive computing, human computer interaction, network multimedia and elearning technology. Prof.Shi is the Chairperson of the Technique Committee of Pervasive Computing, China Computer Federation. Prof.Shi has been guiding the Smart Space, Multimodal Interface, Pervasive Computing and Network Multimedia projects.
Monica Lam
Open Mobile and Social Computing
Abstract
Being with us all the time, mobile devices will revolutionize our social life. We believe fundamentally that we should be able to interact with each other, without having to join the same proprietary social network. Our research approach is to create system platforms that (1) attract users by enabling innovative and attractive interaction paradigms and (2) attract developers by hiding the complexity in creating privacy-preserving applications.
This talk introduces three recent research results in this project. We will describe Junction, an infrastructure for ``partyware'' that enables people at a party to exchange information and collaborate easily and privately. To simplify the specification of access control, we have developed an algorithm that automatically infers important groups from one social activities such as email communications and tagged photos. Lastly, we will present Mr. Privacy, an attempt to create an open and federated social networking infrastructure by leveraging email.
Biography
Monica Lam is a Professor in the Computer Science Department at Stanford University. She is currently working on the Programmable Open Mobile Internet (POMI) 2020 project. Her past research projects include parallel architectures, compiler optimizations for locality and parallelism, program analysis to improve security and reliability, and virtualization-based desktop management. She helped found Tensilica, a company that specializes in configurable processor cores, in 1998. She co-founded Moka5 in 2005 with her students to commercialize their research on desktop virtualization. She is a co-author of the second edition of the dragon book, Compilers: Principles, Techniques, and Tools, published in 2006. She received a B.Sc. from University of British Columbia in 1980 and a Ph.D. in Computer Science from Carnegie Mellon University in 1987. She is a Fellow of the ACM.
Robert Lovas
Convergence of distributed computing infrastructures with focus on Desktop
GridsAbstract
Traditional Desktop Grids consist of computers and other devices, including desktop PCs and notebooks that are used for general purposes but having unused computational and storage capacities. Nowadays there are more and more efforts to help harness GPUs, game consoles, clouds, and human intelligence as well.
These distributed infrastructures can be formed inside research institutes and universities (local Desktop Grids) or by citizens that voluntarily donate spare computing time to science (volunteer Desktop Grids). Both types of Grids collect large number of underutilized resources and can offer them for various simulations and other applications. To be more useful for researchers and students, Desktop Grids have been integrated into scientific workflows on a regular basis; the elaborated generic bridge between Desktop Grids and traditional service Grids together with the appropriate application development methodology can foster the convergence of distributed computing infrastructures.
The recently launched DEGISCO project transfers the knowledge concerning this combined distributed computing infrastructure towards new communities by supporting the creation of new Desktop Grids for e-Science worldwide. Moreover, the project members connect the different types of Grids applying the bridge technology, support the production level combined Grid infrastructure, assist in porting applications, as well as disseminate, promote and provide training about the Grid and its usage. Several scientific applications with large user communities are already available but DEGISCO is searching potential new applications that benefit of the research infrastructure.
The International Desktop Grid Federation has been set-up to exchange experience about the usage of Desktop Grid technology, to expand scientific infrastructures, and in order to bring together Grid operators, application developers, and other key players.
The talk will outline the above described efforts, achievementsm and the progress towards convergence.
Biography
Dr. Robert Lovas is the deputy head of the Laboratory of the Parallel and Distributed Systems at MTA SZTAKI - Computer and Automation Research Institute, Hungarian Academy of Sciences. He received his MSc and PhD degrees at the Budapest University of Technology and Economics. He is the coordinator of the DEGISCO international Grid project in the European Union 7th Framework Programme, member of the Technical Committee of the Hungarian Grid Competence Center, expert of the Grid Application Support Center (GASuC) in collaboration with the European Grid Infrastructure (EGI), and former member of the EU CoreGrid Network of Excellence. From 1997 he has been involved in several national (Chemistry Grid, HAGRID, WEB2GRID), intergovernmental bilateral, and European Grid projects (all phases of EGEE, GRIDCOORD, CANCERGRID, ETICS-2, EDGeS). In the HARNESS project he worked as an exchange scientist at the Department of Math and Computer Science, Emory University, Atlanta, USA. He has long-running experience in ICT collaborations with academic organizations, universities, and enterprises from various application areas of meteorology, biotechnology, telecommunication, and computational chemistry. He is a co-author of more than 40 scientific papers on parallel software engineering and Grid computing particularly from design, debugging, and application aspects.
