1. 1 Energy-efficient Localization Via Personal Mobility Profiling Ionut Constandache Co-authors: Shravan Gaonkar, Matt Sayler, Romit Roy Choudhury and Landon Cox

2. 2 Context Pervasive wireless connectivity + Localization technology = Location-based applications (LBAs)

3. 3 Context Pervasive wireless connectivity + Localization technology = (iPhone AppStore: 3000 LBAs, Android: 600 LBAs) Location-based applications (LBAs)

4. 4 Location-Based Applications (LBAs) Two kinds of LBAs:  One-time location information: Geo-tagging, location-based recommendations, etc.

5. 5 Location-Based Applications (LBAs) Two kinds of LBAs:  One-time location information: Geo-tagging, location-based recommendations, etc.  Localization over long periods of time: GeoLife: shopping list when near a grocery store TrafficSense: real-time traffic conditions

6. 6 Localization Technology LBAs rely on localization technology to get user position

7. 7 Localization Technology LBAs rely on localization technology to get user position Accuracy Technology 10m GPS 20-40m WiFi 200-400m GSM Accuracy Technology 10m GPS 20-40m WiFi 200-400m GSM

8. 8 Localization Technology LBAs rely on localization technology to get user position LBAs executed on mobile phones Accuracy Technology 10m GPS 20-40m WiFi 200-400m GSM Accuracy Technology 10m GPS 20-40m WiFi 200-400m GSM

9. 9 Localization Technology LBAs rely on localization technology to get user position LBAs executed on mobile phones Accuracy Technology 10m GPS 20-40m WiFi 200-400m GSM Accuracy Technology 10m GPS 20-40m WiFi 200-400m GSM Energy Efficiency is important (localization for long time)

10. 10 Localization Technology Ideally Accurate and Energy-Efficient Localization Ideally Accurate and Energy-Efficient Localization

11. 11 Energy … sample every 30s Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h

12. 12 Energy … sample every 30s Battery shared with  Talk time, web browsing, photos, SMS, etc. Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h

13. 13 Energy … sample every 30s Battery shared with  Talk time, web browsing, photos, SMS, etc. Localization energy budget only percentage of battery  20% of battery = 2h GPS or 8h WiFi Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h

14. 14 Energy … sample every 30s Battery shared with  Talk time, web browsing, photos, SMS, etc. Localization energy budget only percentage of battery  20% of battery = 2h GPS or 8h WiFi Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h Battery Lifetime: GPS ~ 10h WiFi ~ 40h GSM ~ 60h For limited energy budget what accuracy to expect?

15. 15 L(t 0 ) L(t 1 ) L(t 2 ) L(t 3 ) L(t 4 ) L(t 6 ) L(t 7 ) L(t 5 ) Problem Formulation

16. 16 L(t 0 ) L(t 1 ) L(t 2 ) L(t 3 ) L(t 4 ) L(t 6 ) L(t 7 ) L(t 5 ) Localization Error t0t0 t1t1 t2t2 t3t3 t4t4 t5t5 t6t6 t7t7 Time Problem Formulation

17. 17 L(t 0 ) L(t 1 ) L(t 2 ) L(t 3 ) L(t 4 ) L(t 6 ) L(t 7 ) L(t 5 ) Localization Error t0t0 t1t1 t2t2 t3t3 t4t4 t5t5 t6t6 t7t7 Time GPS Problem Formulation

18. 18 L(t 0 ) L(t 1 ) L(t 2 ) L(t 3 ) L(t 4 ) L(t 6 ) L(t 7 ) L(t 5 ) Localization Error t0t0 t1t1 t2t2 t3t3 t4t4 t5t5 t6t6 t7t7 Time GPS Problem Formulation Accuracy gain from GPS Eng.: 1 GPS read

19. 19 L(t 0 ) L(t 1 ) L(t 2 ) L(t 3 ) L(t 4 ) L(t 6 ) L(t 7 ) L(t 5 ) Localization Error t0t0 t1t1 t2t2 t3t3 t4t4 t5t5 t6t6 t7t7 Time GPS Accuracy gain from GPS Eng.: 1 GPS read Problem Formulation Accuracy gain from WiFi Eng.: 1 WiFi read WiFi

20. 20 L(t 0 ) L(t 1 ) L(t 2 ) L(t 3 ) L(t 4 ) L(t 6 ) L(t 7 ) L(t 5 ) Localization Error t0t0 t1t1 t2t2 t3t3 t4t4 t5t5 t6t6 t7t7 Time GPS Accuracy gain from GPS Eng.: 1 GPS read Problem Formulation Accuracy gain from WiFi Eng.: 1 WiFi read WiFi

21. 21 Given energy budget B, known Trace T, location readings costs e gps, e wifi, e gsm : Schedule location readings to minimize Average Localization Error (ALE) Given energy budget B, known Trace T, location readings costs e gps, e wifi, e gsm : Schedule location readings to minimize Average Localization Error (ALE) Problem Formulation

22. 22 Given energy budget B, known Trace T, location readings costs e gps, e wifi, e gsm : Schedule location readings to minimize Average Localization Error (ALE) Given energy budget B, known Trace T, location readings costs e gps, e wifi, e gsm : Schedule location readings to minimize Average Localization Error (ALE) Problem Formulation ALE = Avg. dist. between reported and actual location of the user

23. 23 Given energy budget B, known Trace T, location readings costs e gps, e wifi, e gsm : Schedule location readings to minimize Average Localization Error (ALE) Given energy budget B, known Trace T, location readings costs e gps, e wifi, e gsm : Schedule location readings to minimize Average Localization Error (ALE) Problem Formulation ALE = Avg. dist. between reported and actual location of the user Find the Offline Optimal Accuracy

24. 24 Results B = 25% Battery Opt. GPS/WiFi/GSM Trace 178.5m Trace 258.6m Trace 362.1m

25. 25 B = 25% Battery Opt. GPS/WiFi/GSM Trace 178.5m Trace 258.6m Trace 362.1m Offline Optimal ALE > 60m Results

26. 26 Offline Optimal ALE > 60m Results Online Schemes Naturally Worse B = 25% Battery Opt. GPS/WiFi/GSM Trace 178.5m Trace 258.6m Trace 362.1m

27. 27 Our Approach: EnLoc Reporting last sampled location increases inaccuracy

28. 28 Our Approach: EnLoc Reporting last sampled location increases inaccuracy Prediction opportunities exist  Exploit habitual paths  Leverage population statistics when the user has deviated

29. 29 Our Approach: EnLoc Reporting last sampled location increases inaccuracy Prediction opportunities exist  Exploit habitual paths  Leverage population statistics when the user has deviated EnLoc Solution:  Predict user location when not sampling  Sample when prediction is unreliable

30. 30 EnLoc: Overview Deviations EnLoc Habitual Paths E.g. Regular path to officeE.g. Going to a vacation

31. 31 EnLoc: Overview Deviations EnLoc Habitual Paths E.g. Regular path to office Per-user Mobility Profile E.g. Going to a vacation

32. 32 EnLoc: Overview Deviations EnLoc Habitual Paths E.g. Regular path to officeE.g. Going to a vacation Per-user Mobility Profile Population Statistics

33. 33 Profiling Habitual Mobility Intuition: Humans have habitual activities  Going to/from office  Favorite grocery shop, cafeteria

34. 34 Profiling Habitual Mobility Intuition: Humans have habitual activities  Going to/from office  Favorite grocery shop, cafeteria Habitual activities translate into habitual paths  E.g. path from home to office

35. 35 Profiling Habitual Mobility Intuition: Humans have habitual activities  Going to/from office  Favorite grocery shop, cafeteria Habitual activities translate into habitual paths  E.g. path from home to office Habitual paths may branch  E.g., left for office, right for grocery  Q: How to solve uncertainty?  A: Schedule a location reading after the branching point.

36. 36 Per-User Mobility Graph User Habitual Paths Graph of habitual visited GPS coordinates

37. 37 Per-User Mobility Graph User Habitual Paths Logical Representation Graph of habitual visited GPS coordinates

38. 38 Per-User Mobility Graph Graph of habitual visited GPS coordinates  Sample location after branching points  Predict between branching points  # of BPs < # of location samples ( BP = branching point ) User Habitual Paths Logical Representation

39. 39 Evaluation: Habitual Paths 30 days of traces, loc. battery budget 25% per day Assume phone speed known

40. 40 Evaluation: Habitual Paths 30 days of traces, loc. battery budget 25% per day Assume phone speed known

41. 41 Evaluation: Habitual Paths 30 days of traces, loc. battery budget 25% per day Assume phone speed known Average ALE 12m

42. 42 Predict based on population statistics If user on a certain street, at the next intersection predict the most probable turn. Deviations from habitual paths

43. 43 Predict based on population statistics If user on a certain street, at the next intersection predict the most probable turn.  Probability Maps computed from Google Map simulation Deviations from habitual paths

44. 44 Predict based on population statistics If user on a certain street, at the next intersection predict the most probable turn.  Probability Maps computed from Google Map simulation Deviations from habitual paths Goodwin & Green U-TurnStraightRightLeft E on Green00.8810.0390.078 W on Green000.5960.403 N on Goodwin 00.6400.3590 S on Goodwin 00.51300.486

45. 45 Evaluation: Population Statistics OptX: report last sampled location using sensor X (offline) EnLoc-Deviate: Equally spaced GPS + population statistics (online). ALE ~ 32m OptX: report last sampled location using sensor X (offline) EnLoc-Deviate: Equally spaced GPS + population statistics (online). ALE ~ 32m B = 25% Battery

46. 46 Future Work/Limitations Assumed phone speed known  Infer speed using accelerometer  Energy consumption of accelerometer relatively small Deviations from habitual paths  Quickly detect/switch to deviation mode Probability Map hard to build on wider scale  Statistics from transportation departments

47. 47 Conclusions Location is not for free  Phone battery cannot be invested entirely into localization Offline optimal accuracy computed  For specified energy budget  Known mobility trace However, online localization technique necessary EnLoc exploit prediction to reduce energy  Personal Mobility Profiling  Population Statistics

48. 48 Questions? Thank You! Visit the SyNRG research group @ http://synrg.ee.duke.edu/