1 / 10

Self-Adaptation for Everyday Systems

Self-Adaptation for Everyday Systems. Svein Hallsteinsen, Erlend Stav, Jacqueline Floch SINTEF ICT Norway. noise. user context. light. position. system context. battery. shared devices. network QoS. Mobile use means changing context. influence user’s needs.

leo-pace
Download Presentation

Self-Adaptation for Everyday Systems

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Self-Adaptation for Everyday Systems Svein Hallsteinsen, Erlend Stav, Jacqueline Floch SINTEF ICT Norway

  2. noise usercontext light position systemcontext battery shareddevices network QoS Mobile use means changing context influenceuser’s needs • When people are moving around using handheld networked devices, the operating environment for the provided services vary • The user activity and location change, influencing user needs • The capabilities of computers and networks change, influencing the service quality properties • Applications depend upon system resources to satisfy user needs • When resources and needs vary, static applications fail to satisfy user needs most of the time • Changing context requires adaptive applications! mobile user provided quality preferred quality mismatch! application affect operation

  3. Focus • Aspects of self-management we do address • Main: software architecture, mobile computing, algorithms for system adaptation, adaptive components • Aspects we do not address • Prog. language support, AI techniques, autonomic computing, assurance • Targeted application types and properties • Mobile computing used in a changing environment • Everyday systems, for which non-perfect adaptation is acceptable • Cost effective development • Technical contributions • Overall architecture for model based adaptation middleware, combining a system family approach with context awareness and runtime adaptation • Uniform modelling of quality and context properties of components and composites

  4. noise usercontext light architecture model position components nodes systemcontext adaptationmiddleware battery shareddevices network QoS Overall approach to adaptive applications influenceuser’s needs describes relation mobile user monitors preferred quality provided quality selects application variant adaptable application adapts monitors affect operation describes relation

  5. FAMOUS adaptation middleware 1. change event Resource manager Context monitor 2. plan 6. change Adaptation manager 7. change 3. consult 4. plans 5. configure Planner Configurator Overall architecture Application framework FAMOUS middleware Application instance Architecture model Meta model Component repository Component instances FAMOUS component platform Distributed computing environment

  6. Self-reliant client variant Medium client variant Client side Server side Client side Server side o o o UI o UI n n o App logic o App logic Db n o n o Db Dbproxy n n o n Architecture model Application or Medium client Self reliant client and UI App. logic DB Db proxy or UI(hands-free) UI(normal) Applogic Db DB proxy Component repository

  7. Medium client variant avy = if n.nsb>80 then 100 else 100*(1-(80-n.nsb)/80) rsp = if n.nbw>80 then 10 else 10+100*(80-n.nbw)/80) haf = UI.o.haf Client side Server side o o UI n o App logic Db n o n mem = UI.n.mem+Applogic.n.mem nbw =? nsb =? Property annotation of variants haf=yes haf=no o o o o Hand freeUI Normal UI App logic Db c c c c n n n n mem=30 mem=10 mem=10

  8. Utility function • Selection of variant is based on a utility function • User preferences are to decide weight each properties c exe app o n o n usr utility = if n.mem > exe.o.mem then 0 else ( ( (if usr.n.avy <= app.o.avy then 1 else 1- (usr.n.avy-app.o.avy)/usr.n.avy) + (if usr.n.rsp >= app.o.rsp then 1 else 1- (app.o.rsp-usr.n.rsp)/app.o.rsp) + (If ((usr.n.haf and app.o.haf) or (!usr.n.haf and !app.o.haf) then 1 else 0 ) ) / 3 )

  9. Remaining questions • How accurate do we need to model to achieve useful adaptations for everyday systems? • Non-exact models acceptable as long as benefit are provided to the user • Is exhaustive search for best variant too expensive? • Resource constrained devices • Depends on number of variation points in application framework • Current approach focused on structural adaptations of components and connections within an application • How to extend the approach to multiple applications? • How to support other kinds of adaptation (e.g. adjustment of parameters with continuous range in components)?

  10. Organization and projects • SINTEF • Largest independent research organisation in Scandinavia with about 1700 employees • FAMOUS project (2003-2006) • Strategic research project at SINTEF, funded by the Research Council of Norway • Includes a PhD work on context awareness • For more info: www.famous-project.net • MADAM project (2004-2007) • EU-funded research project with industrial and university partners • Additional topics • Modelling tool support • Network level middleware • Industrial pilot applications and experiments

More Related