Distributed data fusion in Distributed data fusion in peer-to-peer environmentpeer-to-peer environment
Sergiy Nazarko, InBCT 3.2, Agora center, University of Jyväskylä
Data fusionData fusion
Branch of applied mathematicsCombines different pieces of information to
receive:– new compatible information– more accurate data
Sundial – simple example of Sundial – simple example of data fusiondata fusion
Data fusion applicationsData fusion applications
Military– target tracking– target identification– data association– situation assessment
Non-military– machine vision– medical decision
support systems– environmental
monitoring
Multisensor data fusionMultisensor data fusion
Improved estimatesProblems:
– corrupt data– different data– different level of precision– conflicting data
Area of interestArea of interestData fusion algorithms which can be
used for target tracking and identification– Transferable Belief Model– Kalman Filtering
““Eye Of Ra”Eye Of Ra”
User InterfaceTBM Kalman Filter
Decentralized data fusion Decentralized data fusion systemssystems
Collection of processing nodesNone of the nodes has knowledge about the
overall network topologyEach node performs a specific computing
taskNo central node exists that controls the
network
Features of DDFSsFeatures of DDFSs
Reliability– no central node– loss of nodes or links does not prevent rest of
the system from functioningFlexibility
– nodes can be added or deleted by making only local changes
– only establishment of links to one or more nodes is needed
Work doneWork done
Master’s thesises: – S. Nazarko, Evaluation of Data Fusion Methods Using
Kalman Filter and TBM– V. Smirnova, Multiagent System for Distributed Data
Fusion in Peer-to-Peer Environment
Gained experience in applying data fusion methods
“Eye Of Ra”
Work in processWork in processIntegration of evaluated algorithm into Chedar
– To get a little bit clearer picture on this step only Kalman filter will be implemented as part of Chedar
Interaction between nodesInteraction between nodes
Network componentsNetwork components
-little Square – sensor node with transmission capabilities
- bold square –control node with sensor’s node capabilities
GUI – user interface which displays tracking trajectory.
Future workFuture work
Further learning of data fusion methodsFusion of TBM and Kalman filterImplementing totally distributed data fusion
system based on peer-to-peer platformEvaluation and research
Thank you!Thank you!