Presentation is loading. Please wait.

Presentation is loading. Please wait.

CAMAS Virtual Machine (CAMAS VM): Multimodal Sensor data stream Querying, Analysis, and Transformation Infrastructure.

Similar presentations


Presentation on theme: "CAMAS Virtual Machine (CAMAS VM): Multimodal Sensor data stream Querying, Analysis, and Transformation Infrastructure."— Presentation transcript:

1 CAMAS Virtual Machine (CAMAS VM): Multimodal Sensor data stream Querying, Analysis, and Transformation Infrastructure

2 CAMASVMResponsphere infrastructure

3

4

5 Computing, visualization, and datasets ● RAID storage server ● multi-tile visualization display ● 8-32-bit-processor IBM server ● 8-64-bit-processor Sun server ● Datasets: drill, 911 calls, 9-11, CAD, GIS, people counter logs, LDC TDT4, disasters, KDD, UCI facilities http://rescue-ibm.calit2.uci.edu/datasets/ CAMASVMResponsphere infrastructure

6 Augmented drill Real drill Simulated drill Emergency response –Situation Awareness (e.g., QUASAR + SAMI) –Customized dissemination Security surveillance –Surveillance –Privacy-preserving surveillance IT testing –Augmenting a drill with virtual people, hazards, and technology CAMASVMResponsphere infrastructure Applications

7 Provide general software infrastructure that allows to Collect, transform, and analyze real-time data from an heterogeneous sensing infrastructure CAMASVMProblem

8 Multi-modality streams –Raw media, structured data, and event streams Abstraction –Abstraction from sensors access protocols, sensors stream data format, synchronization -> high-level language + virtual sensors Processing –Different applications are a formed by intersecting sets of operators: Analysis, transformation, querying, event detection, synchronization, etc -> reuse of operators (code and execution) Scalability –Scalable in terms of number of sensors, volume of data, processing power needs, and network bandwidth availability -> distributed computation + optimization and adaptation Extensibility –Different sensor types, querying and transformation operators, and stream data types -> centralized configuration manager –Failure of nodes and changing availability of processing nodes CAMASVMRequirements

9 Privacy –Obfuscation of sensor data, Privacy policy management and enforcement, Non-derivility by observation Power-awareness –Sensors and processing nodes might be power-limited. –Different operators have different power consumption –Mobility CAMASVMRequirements

10 RunTime Deployer SATLite CQL Other query transf lang... Sensor infrastructure User Heterogeneous sensors Distributed Reflexive Mobile-agent based Deployment optimization Deployment (topology) description High-level query languages Application CAMASVMArchitecture

11 Deployer SATLite CQ L... Sensor RunTim e Processi ng node Processi ng node RunTim e MAMA MAMA Sensor stream MAMA RunTime Infrastructure directory CAMASVM snapshot, configurations, operators repository RunTi me CAMASVMArchitecture User Applicati on MAMA MAMA

12 CAMASVMSATLite cam1 = Create(Cam_humanoid, M1); cam2 = Create(Cam_humanoid); union = Create(Union); alert = Create(Alert, M3); cam_stream1 = Read_cam(C1); cam_stream2 = Read_cam(C2); event_stream1 = cam1(cam_stream1); event_stream2 = cam2(cam_stream2); event_stream3 = union(event_stream1, event_stream2); alert(event_stream3); cam_stream1 event_stream1 (C1)------------->[cam1]-------------------| | cam_stream2 event_stream2 V event_stream3 (C2)------------->[cam2]--------------->[union]-------------->[alert] --------------- MoveTo(cam1, M3); MoveTo(cam2, M3);

13 CAMASVMDeployer Deployer Query XML file Deployed operators Snapshot

14 CAMASVMInfrastructure Directory

15 0.- Sensor: Internet WebCam located at the kitchen. 1.- Data acquisition sensor: gets video stream from sensor and transforms it into a stream of JPEG frames. 2.- Basic event detection: transforms a stream of JPEGs into a stream of events that indicate if the coffee pot is on the burner. 3.- Basic event detection: transforms a stream of JPEGs into a stream of events that indicate the coffee level in the coffee pot (level=0 means coffee pot empty). 4.- Basic event detection: transforms a stream of JPEGs into a stream of events that indicate if the burner is on. 5.- Basic event detection: transforms a stream of JPEGs into a stream of events that indicate if coffee is flowing into the pot. 6.- Temporal synchronizer. 7.- Higher event detection: transforms a stream of synchronized basic events into a stream of events that indicate the risk of the coffee pot exploding: if coffee pot is left empty on the burner with the burner on and no coffee is flowing into the burner. 8.- Visualization: GUI that indicates if there is a risk of the coffee pot exploding. 9.- Visualization: GUI that displays JPEG frames. 7 3 2 61 4 5 08 9 CAMASVMSample Application

16 Privacy-preserving application –Obfuscation of sensor data unless rule is violated (e.g., coffee-pot not in place)

17 CAMAS Virtual Machine (CAMAS VM): Multimodal Sensor data stream Querying, Analysis, and Transformation Infrastructure http://www.ics.uci.edu/~projects/camasvm dmassagu@uci.edu Thank you

18 Extra slides

19 stream MAMA Sensor Deployer SATLite CQ L... Sensor RunTim e Sensor Volunte er machin e Sensor RunTim e MAMA MAMA MAMA MAMA Sensor stream MAMA MAMA MAMA MAMA... RunTime Infrastructure directory CAMASVM snapshot, configurations, operators repository RunTi me CAMASVMArchitecture User Applicati on


Download ppt "CAMAS Virtual Machine (CAMAS VM): Multimodal Sensor data stream Querying, Analysis, and Transformation Infrastructure."

Similar presentations


Ads by Google