DOC PREVIEW
LECTURE NOTES

This preview shows page 1-2 out of 5 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

Chameleon: A Capability Adaptation System for Interface VirtualizationSang-bum Suh1, Xiang Song2, Jatin Kumar2, Dushmanta Mohapatra2,Umakishore Ramachandran2, Jung-Hyun Yo o1, IlPyung Park11SW laboratories, Samsung Electronics, Korea2Georgia Institute of Technology, Atlanta, GA, [email protected], [email protected], [email protected], [email protected]@cc.gatech.edu, [email protected], [email protected] computing capabilities of embedded devices in-crease, services such as Internet and multimedia arecomportably supported by such devices. Users may de-mand for migrating services from one machine to an-other that can provide better capabilities and conve-nience in terms of battery sto ck and input/output (I/O)method. However, I/O devices capabilities between em-bedded mobile devices and user interaction behavior areparticulary different. So, heterogenous I/O capabili-ties and functions may make migration of computingstates between embedded devices virtually impossible.To guarantee seamless I/O state migration between em-bedded mobile devices, we propose interface virtualiza-tion and capability adaptation. In this paper, we de-scribe a system architecture based on Xen virtualizationas well as design principles and performance.1 IntroductionPeople nowadays tend to use their mobile devices toenable their work any place at any time. However, mo-bile devices are usually constraint by its limited power,storage and screen size. People carrying mobile devicesmay want to take advantage of nearby resources in theenvironment to enhance their experience. For exam-ple, a frequent traveller may want to use a high defini-tion TV equipped in an airport lounge to see his moviethat stores in his personal digital assistant (PDA). Ina typical ubiquitous computing environment, these I/Odevices becomes pervasive in any surroundings, whichallows people to connect and use. We need a compre-hensive infrastructure that can enable the connectionbetween people’s mobile devices and the environment.In addition to use the environment provided re-sources, mobile users may also want to seamlessly mi-grate their work from place to place. Taking the aboveexample, if the person who is watching a movie at thelounge doesn’t finish it before boarding, he may wantto “move” his movie to the small monitor in front ofhis seat to continue it without worrying about where heleft off . In this case, his PDA will c oordinate with dif-ferent environments (airport lounge and the airplane)to transfer the internal “state” of the movie player fromone place to the other in order to migrate such a movieplaying service seamlessly.One important issue for such a migration is thatthe source and the destination I/O devices are al-ways heterogenous. Output devices may have differentsizes/resolutions, color settings and even supports forcomplex operations such as 3D. Input devices may alsobe different in the keyboard mapping, mouse accuracyor keypad functions. A good migration system has totake these heterogeneities into consideration and adaptdifferent capabilities of the source and destination de-vices to m ake the migration as seamless as possible.In this pap e r, we will introduce a system, calledChameleon, that can dynamically adapt the capabilitiesof different I/O devices by allowing the dynamic instal-lation of capability adaptors. We provide transparentadaptation to applications with no additional require-ments for the application to be aware of it. However,we are not trying to develop an intriguing capabilityadaptation algorithm. A lot of work in HCI [6][7] isorthogonal to our work and can be complementary toour system for best user experience. Our work is alsoin parallel with application level adaptations such as[1][3], which requires applications to be aware of the de-vice change and therefore can better adjust their UI forthe device change. These projects are also orthogonal toour work and our system can work better if applicationsare not designed to allow such adaptation.The rest of the paper is organized as the following:after presenting our design principles in section 2, webriefly describe the Xen and why we choose Xen as the1base system in section 3. Then, in section 4 we presentour design, followed by details of implementation in sec-tion 5 and the performance evaluation in section 6. Fi-nally, after briefly summarizing other people’s relatedprojects, we conclude our paper and describe p os siblefuture work.2 Design principlesSince we consider the migrations of people’s activitiesacross environments, in our system, there is always amobile platform (guest) that is moved with user and astationary environment system (host) that can provideresources locally to user. The guest may be migratedto different host systems if user moves, where it mayneed to migrate all its previous states to a new host andcontinue its activities by using new resources providedby the new host.The first design principle is that we want to keep theguest unchanged and make the host adapt to the guest.Since in a typical ubiquitous environment, the resourcesare always heterogeneous and therefore the adaptationhas to happen in order to enable the seamless migra-tion. Instead of making the guest adapt to the host, itis always better for the host to adapt the guest since thehost always knows the local resources better and there-fore understand how to adapt a generic guest using itslocal resources.The second design principle is that we want to builda system that can support the dynamic installing anduninstalling of different capability adaptors for dynamicadaptation. The adaptation algorithms, however, arenot our main concern and are out of the scope of thissystem. We rely on other researchers (probably in HCI)to develop the algorithm for adaptation.The third design principle is that we want to makethe selec tion of adaptation algorithms automatic and atrun-time. We notice that there are usually a variety ofadaptation algorithms that can be supported and choos-ing one algorithm over the other is always a challenge.A successful system should support the selection of analgorithm when the actual requirements come.The fourth design principle is that we want to makethe migration as seamless as possible. Therefore, wemay need to migrate the device states (such as framebuffer for display devices ) along with the domain mi-gration. By dumping and resuming the device states,we can ensure the seamless migration


LECTURE NOTES

Download LECTURE NOTES
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view LECTURE NOTES and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view LECTURE NOTES 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?