Hot Topics in Mobile and Pervasive Computing Discussion of LOC1 and LOC2
اسلاید 1: 1CSCE 5013: Hot Topics in Mobile and Pervasive ComputingDiscussion of LOC1 and LOC2Nilanjan BanerjeeHot Topic in Mobile and Pervasive ComputingUniversity of ArkansasFayetteville, ARnilanjan.banerjee@gmail.comAcknowledgment: Romit Roychoudhuri for the slides
اسلاید 2: 2LOC2: SurroundSense
اسلاید 3: Location-Based Applications (LBAs)For Example:GeoLife shows grocery list when near WalmartMicroBlog queries users at a museumLocation-based ad: Phone gets coupon at StarbucksiPhone AppStore: 3000 LBAs, Android: 500 LBAs
اسلاید 4: Most emerging location based apps do not care about the physical locationGPS: Latitude, Longitude
اسلاید 5: Most emerging location based apps do not care about the physical locationInstead, they need the user’s logical locationGPS: Latitude, LongitudeStarbucks, RadioShack, Museum, Library
اسلاید 6: Physical Vs LogicalUnfortunately, most existing solutions are physicalGPSGSM basedGoogle LatitudeRADARCricket…
اسلاید 7: Given this rich literature,Why not convert from Physical to Logical Locations?
اسلاید 8: Physical LocationError
اسلاید 9: Pizza HutStarbucksPhysical LocationError
اسلاید 10: Pizza HutStarbucksPhysical LocationError The dividing-wall problem
اسلاید 11: SurroundSense:A Logical Localization Solution
اسلاید 12: It is possible to localize phones by sensing the ambience Hypothesissuch as sound, light, color, movement, WiFi …
اسلاید 13: It is possible to localize phones by sensing the ambience Hypothesissuch as sound, light, color, movement, WiFi …
اسلاید 14: Multi-dimensional sensing extracts more ambient informationAny one dimension may not be unique, but put together, they may provide a unique fingerprint
اسلاید 15: SurroundSenseMulti-dimensional fingerprintBased on ambient sound/light/color/movement/WiFi StarbucksWallPizza Hut
اسلاید 16: BACDEShould Ambiences be Unique Worldwide?FGHJILMNOPQQRK
اسلاید 17: Should Ambiences be Unique Worldwide?BACDEFGHJIKLMNOPQQRGSM provides macro location (strip mall) SurroundSense refines to Starbucks
اسلاید 18: Economics forces nearby businesses to be diverseNot profitable to have 3 adjascent coffee shopswith same lighting, music, color, layout, etc.SurroundSense exploits this ambience diversityWhy does it work?The Intuition:
اسلاید 19: +Ambience FingerprintingTest FingerprintSound Acc.Color/LightWiFiLogicalLocation MatchingFingerprintDatabase=Candidate FingerprintsGSM Macro LocationSurroundSense Architecture
اسلاید 20: FingerprintsSound:(via phone microphone)Color:(via phone camera) Amplitude Values-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 Normalized Count0.14 0.12 0.1 0.080.06 0.04 0.02 0 Acoustic fingerprint (amplitude distribution)Color and light fingerprints on HSL space Lightness 1 0.5 0 Hue00.5100.20.40.60.81Saturation
اسلاید 21: Fingerprinting Sound Fingerprint generation : Signal amplitudeAmplitude values divided in 100 equal intervalsSound Fingerprint = 100 normalized values valueX = # of samples in interval x / total # of samplesFilter Metric: Euclidean distanceDiscard candidate fingerprint if metric > threshold гThreshold г Multiple 1 minute recordings at the same locationdi = max dist ( any two recordings )г = max ( di of candidate locations )
اسلاید 22: Fingerprinting ColorFloor PicturesRich diversity across different locationsUniformity at the same locationFingerprint generation: pictures in HSL spaceK-means clustering algorithmCluster’s centers + sizesRanking metric
اسلاید 23: FingerprintsMovement: (via phone accelerometer)CafeteriaClothes StoreGrocery StoreStaticMoving
اسلاید 24: FingerprintsMovement: (via phone accelerometer)CafeteriaClothes StoreGrocery StoreStaticQueuing Seated Moving
اسلاید 25: FingerprintsMovement: (via phone accelerometer)CafeteriaClothes StoreGrocery StoreStaticPause for product browsingShort walks between product browsingMoving
اسلاید 26: FingerprintsMovement: (via phone accelerometer)CafeteriaClothes StoreGrocery StoreStaticWalk moreQuicker stopsMoving
اسلاید 27: FingerprintsMovement: (via phone accelerometer)WiFi: (via phone wireless card)CafeteriaClothes StoreGrocery StoreStaticƒ(overheard WiFi APs)Moving
اسلاید 28: Fingerprinting WiFiFingerprint generation: fraction of time each unique address was overheardFilter/Ranking MetricDiscard candidate fingerprints which do not have similar MAC frequencies
اسلاید 29: DiscussionTime varying ambienceCollect ambience fingerprints over different time windowsWhat if phones are in pockets?Use sound/WiFi/movementOpportunistically take picturesFingerprint DatabaseWar-sensing
نقد و بررسی ها
هیچ نظری برای این پاورپوینت نوشته نشده است.