MOBISYS 2011 MOBISYS 2011 The 9th International Conference on Mobile System, Applications, and Services Tracing a Missing Mobile Phone using Daily Observations Hyojeong Shin*, Yohan Chon, Kwanghyo Park, Hojung Cha Hyojeong Shin (hjshin@cs.yonsei.ac.kr) Mobile Embedded System Lab. Yonsei University Hyojeong Shin (hjshin@cs.yonsei.ac.kr) Mobile Embedded System Lab. Yonsei University
Motivation (1) Motivation (1) ( ) ( ) A mobile device is a precious one. A mobile device is a precious one. The device carries personal information The device carries personal information The device carries personal information. The device carries personal information. Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Motivation (2) Motivation (2) ( ) ( ) • A device will be unexpectedly lost . So … • Service Coverage is critical. – A service should cover Indoor space. • GPS coverage : only 20 % Place coverage [*] – A service should employ well-deployed infrastructure. – A service should consider various target devices. • Smartphones, feature phones, laptops, tablets at others • Accuracy: Indoor search requires room-level accuracy. • Also, the service should consider multiple-story building. p y g • Energy: A mobile device has limited power. • Searching Time: A lost device should be found in very short time • Searching Time: A lost device should be found in very short time. [*] Y. Chon, et. al., "Autonomous Management of Personalized Location Provider for Mobile Services," IEEE T SMC-C 2011 Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Motivation (3) Motivation (3) ( ) ( ) • Our answer is employing Wi-Fi fingerprints. SSID RSSI 1 AP1 -60dBm – Well deployed in indoor space. 2 AP2 -90dBm 3 AP3 -45dBm – Well deployed in diverse mobile device. 4 AP4 -77dBm – Wi-Fi fingerprints are unique and stable for space and time. • Problems – Generally, an indoor floor plan is not available. – Wi-Fi fingerprint does not carry location information. g y • FindingMiMo – In our daily life, a device records Wi-Fi fingerprints in daily basis. In our daily life, a device records Wi Fi fingerprints in daily basis. – When it is lost, a chaser application searches the missing device by tracing the series of Wi-Fi fingerprints. g g p Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Related Work Related Work Limitations • Find-A-Lost-Phone Services – Apple Mobile Me & MS Window Lives : Service Coverage – Localization service (WPS) • Infrastructure-based positioning approach : Supervised Area p – Bluetooth-tag, RFID-tag, Ubisense (UWB), g, g, ( ), : Installation Cost Indoor GPS • Mobile-based positioning approach – Place Learning : SensLoc, iLoc, SoundSense, : Training Phase is required. : Energy Cost : Energy Cost SurroundSense and Jigsaw SurroundSense and Jigsaw – Geometric localization(Radio): Radar, PlaceLab – Geometric localization(Sensor): MEMS, Greefield Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
FindingMiMo FindingMiMo Architecture g Architecture Missing-Mobile Part g Chaser Part Missing ‐ Mobile Chaser Missing-Mobile Chaser Installed on an extra device. I t ll d t d i Daily basis Wi-Fi logging (old smartphone, laptop, tablets…) Background service Device tracking application. Low energy consumption Wi-Fi, INS, GPS enabled. Wi Fi INS GPS enabled LifeMap SmartSLAM SmartSLAM Pedestrian Tracking User Context Monitor Constructing a floor plan Place Learning Movement Tracking Movement Tracking *Y. Chon,et.al.,"LifeMap: A Smartphone-based Context Provider for Location-based Service", IEEE Pervasive Comp.2011 **H. Shin, et. al. "SmartSLAM: Constructing an Indoor Floor Plan using Smartphone" Yonsei University, 2011 Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Ambient Log Ambient Log • Inertial sensor is not available. (Energy issue, device diversity) • Log: GPS(Long./Lat.), APs, Status, Accuracy, POI label, timestamp outdoor outdoor indoor indoor SSID RSSI ( ( x , y ) ) 1 AP1 -60dBm Ambient Log A bi L (? ?) (?,?) 2 AP2 -90dBm 3 AP3 -45dBm The log does not reveal the exact 4 4 AP4 AP4 -77dBm 77dB location of the missing device. Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
FindingMiMo FindingMiMo Scenario g Scenario Cold Warm progress vicinity vicinity • The ambient log does not contain the location information. – Analyzing the log, the chaser application displays “warm/cold” signs. y g g, pp p y g – The system signs out “ warm ” when a chaser is headed in the right direction and “ cold ” when he is not, while he is searching. – cf. Warm/Cold game, a treasure hunt game for children f W /C ld h f h ld [Thank you for a reviewer’s comment] Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Design Issue Design Issue g Missing-Mobile Part g Chaser Part • No energy issue (rechargeable) Energy Limitation • – In normal operation: Signal Processing • Energy efficient logging – How to generate the tracking – In missing: – In missing: i f information i NOT rechargeable – Wi-Fi similarity ( warm/cold ) – Adaptive sensing schedule – Searching Progress Searching Progress Storage Limitation • Searching a device • – Storage complexity – How to search to the device – Massive APs are observed. – Warm/cold game – Linear-scale growth is not feasible. – User Interface – Log reduction method – Log reduction method Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Missing Missing-Mobile : Energy (1) g Mobile : Energy (1) gy ( ) gy ( ) • Adaptive Sensing Scheduling p g g – Moving-State monitoring (w/o sensor) • When radio signal becomes stable: stationary state g y • When radio signal rapidly changes: moving state – GPS : GPS : • Turns off GPS in stationary state – Wi-Fi: Wi Fi: • Increases sample interval in stationary state Ti Time Coverage C [*] [*] Time Coverage Ti C [Fi di [FindingMiMo] MiM ] GPS GSM Wi-Fi move stay Coverage Coverage 4 5% 4.5% 99.6% 99 6% 94.5 % 94 5 % Coverage Coverage 13% 13% 87 % 87 % [*] LaMarca, A.a et al., “Place lab: Device positioning using radio beacons in the wild”, pervasive 2005 Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Missing Missing-Mobile : Energy (2) g Mobile : Energy (2) gy ( ) gy ( ) Average energy consumption Average movement time vs. stationary time – People spent approximately 13 % of a day to move. – Adaptive Sensing: 3.7kJ (vs. Continuous sensing : 41.1 kJ [*] ) • reducing the battery’s lifetime by 14% in average. • The result depends on individual usage patterns. [*] D. H. Kim, et. al., SensLoc: Sensing Everyday Places and Paths using Less Energy, Sensys 2010. Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Missing-Mobile : Storage (1) Missing g Mobile : Storage (1) g g ( ) ( ) • Storage Complexity for Logging – The missing-mobile periodically scan GPS and Wi-Fi. h b l d ll d – The log tends to grow in linear scale. – When a user visit known place, the log becomes useless. • c.f. LifeMap provides user’s POI (Point of Interest) and GPS – The redundant log is flushed. – Log: GPS(Long./Lat.), Status, Accuracy, POI label, timestamp, APs Redundant Log Ambient Log Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Missing Missing-Mobile : Storage (2) g Mobile : Storage (2) g g ( ) ( ) When a user visit a known place, the log is flushed the log is flushed. Mbytes) 5 4 d Storage (M 3 2 1 Used 0 6 8 10 12 14 16 18 20 22 24 Time (Hour) • Storage Complexity – All scan data : 4.5 MB to 22.3 MB in a day (1,500 APs) All scan data 4 5 MB to 22 3 MB in a day (1 500 APs) – The storage usage was around 5 MB . (empirical data) Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Chaser : Signal Processing Chaser : Signal Processing g g g g Missing-mobile Missing mobile chaser SSID RSSI 1 AP1 -60dBm 2 2 AP2 AP2 -90dBm 90dB 3 AP3 -45dBm • Wi-Fi signal comparison 4 AP4 -77dBm • Si il it T i • Similarity: Tanimoto Coefficient t C ffi i t • Warm/cold: the similarity of the best-match observation Progress: the index of the best match observation in the log Progress: the index of the best-match observation in the log • • Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Chaser : Searching a Device Chaser : Searching a Device g G E F F D D G G A A larity D simil B C C C E B A time(sec) • Chasing Strategy • Chasing Strategy – Visit a unvisited place in vicinity – Warm : Visit a next place Wa : p – Cold : Come back to the previous place / and visit another place. Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Chaser GUI Chaser GUI Mobile Embedded System Lab. Yonsei University Mobile Embedded System Lab. Yonsei University
Recommend
More recommend
