slide1 n.
Skip this Video
Loading SlideShow in 5 Seconds..
Prof . Congduc Pham http://www.univ-pau.fr/ ~cpham Université de Pau, France PowerPoint Presentation
Download Presentation
Prof . Congduc Pham http://www.univ-pau.fr/ ~cpham Université de Pau, France

Loading in 2 Seconds...

  share
play fullscreen
1 / 22
Download Presentation

Prof . Congduc Pham http://www.univ-pau.fr/ ~cpham Université de Pau, France - PowerPoint PPT Presentation

seven
174 Views
Download Presentation

Prof . Congduc Pham http://www.univ-pau.fr/ ~cpham Université de Pau, France

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Network lifetime and stealth time of wirelessvideosensor intrusion detectionsystemsunderrisk-basedscheduling Prof. Congduc Pham http://www.univ-pau.fr/~cpham Université de Pau, France ISWPC, 2011 Hong-Kong Wednesday, February 23rd

  2. Wireless VideoSensors Imote2 Multimedia board

  3. Surveillance scenario (1) • Randomlydeployedvideosensors • Not onlybarriercoverage but general intrusion detection • Most of the time, network in so-calledhibernate mode • Most of active sensornodes in idle modewithlow capture speed • Sentrynodeswithhigher capture speed to quicklydetect intrusions

  4. Surveillance scenario (2) • Nodesdetecting intrusion must alert the rest of the network • 1-hop to k-hopalert • Network in so-calledalerted mode • Capture speed must beincreased • Ressources shouldbefocused on makingtracking of intruderseasier

  5. Surveillance scenario (3) • Network should go back to hibernate mode • Nodes on the intrusion path must keep a high capture speed • Sentrynodeswithhigher capture speed to quicklydetect intrusions

  6. Wholeunderstanding of the sceneiswrong!!! Don’t miss important events! Real scene Whatiscaptured

  7. How to meet surveillance app’s criticality • Capture speed canbe a « quality » parameter • Capture speed for node v shoulddepend on the app’scriticality and on the level of redundancy for node v • Note thatcapturing an image does not meantransmittingit • V’scapture speed canincreasewhen as V has more nodescoveringitsownFoV - coverset

  8. RedundancyNode’scover set • Each node v has a Field of View, FoVv • Coi(v) = set of nodes v’ such as v’Coi(v)FoVv’ covers FoVv • Co(v)= set of Coi(v) V2 V1 V4 V V3 Co(v)={V1,V2,V3,V4}

  9. Criticality model (1) • Link the capture rate to the size of the coverset • High criticality • Convexshape • Most projections of x are close to the max capture speed • Lowcriticality • Concave shape • Most projections of x are close to the min capture speed • Concave and convexshapesautomaticallydefinesentrynodes in the network

  10. Criticality model (2) • r0canvary in [0,1] • BehaViorfunctions (BV) defines the capture speed according to r0 • r0 < 0.5 • Concave shape BV • r0 > 0.5 • Convexshape BV • We propose to use Beziercurves to model BV functions

  11. Some typical capture speed • Set maximum capture speed: 6fps or 12fps for instance • Nodeswithcoverset size greaterthan N capture at the maximum speed N=6 P2(6,6) N=12 P2(12,3)

  12. How to build an intrusion detection system • Static • Prior to deployment, define r° in [0,1] according to the application’scriticality • Risk-based • R0is set initiallylow : R°min • Somenodes serve as sentrynodes • On intrusion, increase R° to R°maxduring an givenalertperiod (Ta) • After Ta, go back to R°min • 2 variants • R°moves fromR°minto R°max in one step • R°moves fromR°minto R°maxby reinforcementbehavior

  13. Risk-basedscheduling in images (1) • R°=R°min=0.1, R°max=0.9, no alert |Co(vj)|=3 fps=0.14 0.81 |Co(vi)|=8 0.14 fps=0.81 3 8

  14. Risk-basedscheduling in images (2) • R°R°=R°max=0.9 |Co(vj)|=3 Fps=1.9 2.8 1.9 |Co(vi)|=8 Fps=2.8 3 8

  15. Simulation settings • OMNET++ simulation model • Videonodes have communication range of 30m and depth of view of 25m, AoVis 36°. 150 sensors in an 75m.75m area. • Battery has 100 units, 1 image = 1 unit of batteryconsumed. • Max capture rate is 3fps. 12 levels of cover set. • Full coverageisdefined as the regioninitiallycoveredwhen all nodes are active

  16. meanstealthtime (MST) t1-t0is the intruder’sstealth time velocityis set to 5m/s t0 t1 • intrusions startsat t=10s • when an intruderisseen, computes the stealth time, and starts a new intrusion until end of simulation

  17. meanstealthtimestaticscheduling

  18. meanstealth timerisk-basedscheduling 450s 1300s • Sensornodesstartat 0.1 thenincrease to 0.9 if alerted (by intruders or neighbors) and stayalerted for Ta seconds

  19. meanstealth timew/woreinforcement Ir=0.6 • On alert 0.1Ir, then • 2 alertmsg IrIr+1 • Until Ir=R°max • Reinforcementalwaysincreases the network lifetime • Meanstealth time is close to the no-reinforcement case, especiallywhen Ta>20s

  20. Withreinforcementvarious initial threshold • Ir=0.4 • Ir=0.5 • Ir=0.6 • Reduce Iralwaysincreases the network lifetime • For small value of Ta, MST increaseisnoticeable • It isbetter to increase Irthanincrease Ta.

  21. 0 0 <5 <5 <10 <10 <15 <15 >15 >15 Sentry nodes # of cover sets # intrusion detected

  22. Conclusions • Models the application’scriticality as beziercurves and schedules the videonode capture rate according to the redundancylevel • Withthis model, a risk-basedschedulingcanincrease the network lifetimewhilemaintaining a highlevel of service (meanstealth time) • Reinforcementbehaviorisbeneficial and itisbetter to keep the alertperiodlow <=20s for instance