1 / 18

CoCoDa - Data Management in Mobile Ad-Hoc Networks

Joos Böse, Katharina Hahn, Manuel Scholz, Heinz Schweppe http://www.inf.fu-berlin.de/inst/ag-db/projects/cocoda Databases and Information Systems Group Freie Universität Berlin. CoCoDa - Data Management in Mobile Ad-Hoc Networks. Research topic: guarantees for

heulwen-evans
Télécharger la présentation

CoCoDa - Data Management in Mobile Ad-Hoc Networks

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. Joos Böse, Katharina Hahn, Manuel Scholz, Heinz Schweppe http://www.inf.fu-berlin.de/inst/ag-db/projects/cocoda Databases and Information Systems Group Freie Universität Berlin CoCoDa -Data Management in Mobile Ad-Hoc Networks

  2. Research topic: guarantees for computations in unstable environments • Distributed transactions in mobile environments? • Most are client/server oriented • MANETs: Atomic commit more important than isolation • Replication • Information freshness • Conflicting operations • Consistency of local updates? • Application Scenario • E-Commerce • Cooperative documents (e.g. mobile WiKi)

  3. CoCoDa System Architecture MANET Info TA Processor Replication SLS Core Engine Data Retrieval DataDissemination Communication Interface J2ME CDC OSGI JINI Ad-Hoc Wireless Connectivity (IEEE 802.11, Bluetooth)

  4. Replication in MANETs • Goal: Manage distributed state on mobile nodes (update anywhere) • Need for Data Dissemination • Reach as many nodes as possible • Minimize costs for dissemination • Consistent Update • Transactional consistency • Optimistic synchronization • Minimize cost

  5. Fully replicated system • Transactions only access local resources (not distributed TA) • Local pre-comitt • TAlocal (op1,…,opn) is sent to all replica • Order of TA excecution defined by time stamps • Different conflict resolution strategies • Syntactic (e.g. time), not semantic conflict resolution

  6. Dependencies of TAs Conflictingtransactions: both updated the same value established by TA12 timestamp

  7. Dependencies and conflict resolution a) and b): two conflict resolutions "value lineage": b(a), c(a), d(b,a), e(b,a) product byconflicting transaction

  8. Guarantees in unstable systems?? Are we solving the right problems?

  9. Heathland experiments • Field test of sensor network by Thurau (TU HH) • 24 nodes, 16 days, outside in the forestgoal: analysis of algorithms under realistic conditions • Many interesting results, one of them: Loss of packets significant, in particular in case of large multi-hop depth

  10. Heathland (Thurau) success rate  100*0.8depth

  11. Do not try to avoid uncertainty but make the best out of it!

  12. The Home, Year 2020 • 1000s of sensors (light, temp, sound, motion, location, …) • 100s of actuators (locks, switches, heating, water, …) • Masses of data (“Traditional” data plus sensor streams) • Does a lot for you: security, HVAC, energy management & demand-response, entertainment, … • Use Alice’s motion patterns to activate electrical devices (e.g., water heating) • Correlate user motions with existing patterns to detect “suspicious” behavior Intel's vision Berkeley Lab, HeisenData project

  13. Smart home must... Handle uncertainty and correlation ( • P( sensor 2455 fired accurately ) > .8 • P( someone in den | behavior of sensors ) > .95 • ... A hierarchy of inferences, from minute to abstract Recognize, manage, and exploit correlations (spatial, temporal)

  14. Approaches today • Interesting processing done outside DB • Lose key benefits of DBMS (declarative processing, persistence, optimization)

  15. Location... ... is always inexact Uncertain regions of moving objects from B. Yu: A spatiotemporal Uncertainty Model, SAC 06

  16. So what ? Uncertainty of data: a principle problem of mobile data mangement and radio communication • What is needed? Mechanisms dealing with uncertainty at all levels • Adaptation of algorithms to hostile environments • Management of data lineage • Graceful degradation of systems depending on degree of uncertainty • ....

  17. Example: Unreliable RFID sensors Integrity constraints for sensor data cleaning Library example (D. Suciu et al. 06):FOR ALL Sightings S CHECK COUNT(SELECT * FROM SIGHTING T WHERE T.l = S.t AND T.bid = S.bid) <= 1 " A particular book exist only once in the library " Other constraints: " A checked-out book cannot appear in the library " ...

  18. Are we solving the right problems? • to some extent: Yes sensors, data streaming, ... • No? • MANETs are not really needed when it is easy to access the fixed net through access points • Client - Server mobile interaction will dominate • Integrating methods from uncertainty management (ML, probabilistic db,..) into mobile data management seems appropriate

More Related