Using Moving Object Databases to Provide Context Information in - PowerPoint

Document Sample
Using Moving Object Databases to Provide Context Information in - PowerPoint Powered By Docstoc
					Using Moving Object Databases to Provide
      Context Information in Mobile
              Environments
                               Katharina Hahn1
                  A. Voisard1,2, M. Scholz1, H. Schweppe1, J. Böse1

                  1Databases    and Information Systems, FU Berlin
    2Fraunhofer   Institute for Software and Systems Engineering (ISST), Berlin




                                                                                  May 06
  Support Mobile Transactions




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   2   Katharina Hahn
  Overview



        Motivation
        Background
        Related Work
        Context System
               Modeling Context
               Storing Context
               Querying Context
               Exchanging Context
    Conclusion
    Future Work


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   3   Katharina Hahn
  Motivation



    Support transactions in MANETs [1]
                   eLearning scenario
                   Change of electronic goods
                   Distributed commit
            Support atomic commit
                   Replicate recovery information for failing nodes
                   Recovering nodes will follow global decision
                   Provide recovery probability

            How to disseminate recovery information to reach high
             availability?
                   Disseminate among preferably stable nodes
                   Avoid flooding the MANET
            Determine „stable“ nodes within surrounding


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   4   Katharina Hahn
  Background



    System model
            Heterogeneous MANET
            Nodes differ according to
                   Mobility (stable vs. mobile)
                   Resources (highly vs. poorly equipped)


    Context awareness
            Gather knowledge of currently surrounding nodes
            Enable prediction of behavior of nodes
            Consider probability of failure of nodes


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   5   Katharina Hahn
  Related Work



    What is considered to be context?

            Schilit (1994)
                   Three important aspects of context are: where your are, who
                    you are with and what resources are nearby [2].
            Dey (1999)
                   Context is any information that can be used to characterize
                    the situation of an entity [3].


            Too general
                   Mobile peer-to-peer environment with limited resources
                    forces to focus on certain context attributes



Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   6   Katharina Hahn
  Related Work (2)



    Context aware systems
            Numerous approaches exist
            e.g. SOCAM [4], PACE [5]

    Predicting Context
            Not only deals with recognizing and reasoning on
             current context
            Focuses on prediction of context [6]
                     Sensing
                     Feature Extraction
                     Classification
                     Labeling
                     Prediction


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   7   Katharina Hahn
  Overview



        Motivation
        Introduction
        Related Work
        Context System
               Modeling Context
               Storing Context
               Querying Context
               Exchanging Context
    Conclusion
    Future Work


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   8   Katharina Hahn
  Context Information



    Objectives
               MANET system: no central context server
               Light-weight to be run on mobile devices
               Establish local view on MANET (distributed context)
               Enable prediction of future behavior of nodes
               For application use

            Evaluate nodes and their surrounding according to
             their stability
                   Will they be reachable for the duration of the transaction?
                   Will they be reachable by any recovering node so that they
                    can learn about missed steps?



Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   9   Katharina Hahn
  Context Model



    Consider information which indicates
            Connection status
                   Position, movement
                   Link stability, signal quality, transmission range
                   Battery power
            No (external) environmental sensors
             considered
            Do not focus on modeling user-initiated shut-
             down probability


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   10   Katharina Hahn
  Modeling Context Data



    Explore different levels of context
                                                        group context GC
                                                     Owner ID
                                                     {Source ID, context}



                                  local context LC                               auxiliary context
                                 Source ID                                   Owner ID
                                 {attributes}                                {{Source ID}, attribute}
                                                                             e.g. distance


              atomic attribute                           derived attribute                              group level
            {attribute, value}                      {attributes}  attribute                            local level
            e.g. position                           e.g. speed



Managing Context Information and Semantics in Mobile Environments (MCISME)    May 06              11             Katharina Hahn
  Storing Context Data


                                        Every nodes, able to sense its context, can
                                         originate and send local context data
                                        Nodes interested in their surrounding carry local DB
                                        Relational DB with MOD concepts to model
                                         dynamic environment
                                        Dynamic vs. static attributes

    Dynamic attributes analogous to MOST
                                                                                      node_id           N103
     data model [6]
                                                                                      rbt.updatetime t3
             Enables extrapolation (prediction)
                                                                                      rbt.value         00:53:00
             Reduces network load opposed to
              conventional relational DB                                              rbt.function(t)   -(0,0,1)*t
             Uncertainty serves as “contract”                                        rbt.uncertainty   (0,10,0)
             Prediction of behavior is administered within
              local DB

Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06           12               Katharina Hahn
  Querying Context


    Queries are submitted locally
            Conventional querying using SQL
            Temporal queries
                   e.g. querying inter/extrapolated states of dynamic attributes
            Given set of temporal queries implemented
                   e.g. “When will distance be greater than transmission range?”

    Must/may semantic
            Strong indicator for probability that nodes stay connected
            Use degree of overlapping future whereabouts with uncertainty and
             transmission range as foundation for probability evaluation




                            node

                            uncertainty                                               region of interest


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06            13          Katharina Hahn
        Motivation
        Introduction
        Related Work
        Context System
               Modeling Context
               Storing Context
               Querying Context
               Exchanging Context
    Conclusion
    Future Work


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   14   Katharina Hahn
  Exchanging Context



    Before beginning of interaction
    Exchange information within certain number of hops



    Classify involved nodes
            Initiators gather GC
            Sources sense LC
            Inactive nodes




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   15   Katharina Hahn
  Exchanging Context (2)



    Event-triggered
            Explicit update if conventional attribute changes
            Dynamic attributes
                   A.uncertainty serves as contract
                           An update will be sent, whenever the current deviation exceeds
                            the uncertainty
                              |sensed(dynAttr) – calculated(dynAttr)| > dynAttr. uncertainty

                   Adapt uncertainty to nodes behavior, e.g.
                           Sensible value if movement is fast
                           Robust value if remaining battery power is low




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   16       Katharina Hahn
  Exchanging Context (3)



    Time-based
            Update all attributes periodically, if no update
             through event occurred
                   Learn about leaving and newly arrived nodes

                   Data in DB ages if no update of the node is
                    received




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   17   Katharina Hahn
  Context Evaluation



    Evaluation of gathered information using existing
     approaches [8,9]
    Consider combination of energy resources, link
     stability and movement
    Combine through
            Weighted metric                                                           gather context

            Fuzzy logic
                                                                                      evaluate context
            Consider failure probability
                                                                                       evaluate node




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06        18            Katharina Hahn
  Platform


    Implementation
            J2ME CDC
            OSGi
            JAVA database
                                                                                      Context evaluation


    Platforms                                                                  Context                MOD
            Microsoft Windows Mobile                                          Exchange        DB     wrapper
            Navilock NL-216CX Compact Flash


                                                                                                  Sensors




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06       19           Katharina Hahn
        Motivation
        Introduction
        Related Work
        Context System
               Modeling Context
               Storing Context
               Querying Context
               Exchanging Context
    Conclusion
    Future Work


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   20   Katharina Hahn
  Conclusion



    Objectives were
            MANET context model
            Stability evaluation
            Light-weight system
        Multi-level context model
        Storing, querying and exchanging context
        Administrate context prediction within DB
        Part of the CoCoDa middleware that supports
         mobile transactions

Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   21   Katharina Hahn
  Future Work



    Establish different prediction schemes on top of
     context system
    Evaluation of system within middleware
            Additional storage and network cost
            Performance through evaluation of reliable recovery
             log provider

          Does rate of successful recovery increase
           opposed to choosing nodes randomly at
           acceptable costs?


Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   22   Katharina Hahn
                    Thank you for your attention!




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   23   Katharina Hahn
  References


   [1] http://www.inf.fu-berlin.de/inst/ag-db/projects/cocoda/
   [2] Schilit et al.: Context-aware Computing Applications, 1st International Conference on
       Mobile Computing Systems and Applications, 1994.
   [3] Dey, Abowd: Providing Architectural Support for Building Context-Aware Applications, PhD
       Thesis, College of Computing, Georgia Institute of Technology, 2000.
   [4] Gu et. al: A Middleware for Building Context-Aware Mobile Services, n Proceedings of IEEE
       Vehicular Technology Conference, 2004.
   [5] Henricksen, Induslka et al.: Middleware for Distributed Context-Aware Systems, In OTM
       Conferences, 2005.
   [6] Mayrhofer: An Architecture for Context Prediction, Advances in Pervasive Computing, Part
       of 2nd International Conference on Pervasive Computing, 2004.
   [7] Sistla et. al: Modeling and Querying Moving Objects. 13th International Conference on
       Data Engineering, 1997.
   [8] Su, Lee, Gerla: Mobility Prediction In Wireless Networks, International Journal of Network
       Management, 2001.
   [9] Gruenwald and Banik: Power Aware Management Of Mobile Real-Time Database
       Transactions in ad-hoc Networks.




Managing Context Information and Semantics in Mobile Environments (MCISME)   May 06   24   Katharina Hahn