Video and slides synchronized, mp3 and slide download available at URL http://bit.ly/1shcgNu. Mo Ramezanpoor discusses how two different activity tracking apps -"Zombies, Run!" and "The Walk"- have different approaches for tracking user activity. Filmed at qconlondon.com. Mo Ramezanpoor is a software developer working mainly on iOS apps. Last year he joined Six to Start, where he is the lead developer for fitness game Zombies, Run!. Mo has studied Astrophysics and has experience in writing data processing and simulation software.

2. Watch the video with slide
synchronization on InfoQ.com!
http://www.infoq.com/presentations
/sensors-tracking-user-activity
InfoQ.com: News & Community Site
• 750,000 unique visitors/month
• Published in 4 languages (English, Chinese, Japanese and Brazilian
Portuguese)
• Post content from our QCon conferences
• News 15-20 / week
• Articles 3-4 / week
• Presentations (videos) 12-15 / week
• Interviews 2-3 / week
• Books 1 / month

3. Presented at QCon London
www.qconlondon.com
Purpose of QCon
- to empower software development by facilitating the spread of
knowledge and innovation
Strategy
- practitioner-driven conference designed for YOU: influencers of
change and innovation in your teams
- speakers and topics driving the evolution and innovation
- connecting and catalyzing the influencers and innovators
Highlights
- attended by more than 12,000 delegates since 2007
- held in 9 cities worldwide

4. Sensors aren’t enough
Mo Ramezanpoor
Six to Start
@mohsenr

5. Agenda
A look at fitness tracking apps!
• Tailored data processing
• Adapting the UI
Overcoming device limitations!
• Reducing battery use
• Keeping the app alive
Handling sensor data!
• Keeping track of time
• Distance calculation

7. Zombies, Run!

8. The Walk

9. Reducing battery use

10. Reducing battery use
Don’t use CPU if you can!
• Get data points in batches
• Don’t use standalone timers to update the UI
• Defer data processing as much as possible
Only ask for the accurate that you need!
• Accurate location data will require multiple antennae
• Accurate motion data will require multiple sensors
• Accurate data points are computationally more expensive to produce

11. Keeping the app alive (iOS)

12. Keeping the app alive (iOS)
Play audio in the background!
• If you already need to play audio in the background
• If you need to completely stop the device from sleeping
Track location!
• Using significant location change updates
• Using explicit location tracking but with low accuracy
Use notifications!
• If your app’s behaviour depend on external sources
• f you have can assume network connectivity

13. Keeping track of time

14. “Absolute” time
26 35 44…ticks since first tick!

15. “Absolute” time
26 35 44…ticks since first tick!
“System” time
182 186 190…seconds since reference

16. “Absolute” time
26 35 44…ticks since first tick!
“System” time
182 186 190…seconds since reference
“Calendar” time
13:30:08 13:30:12 13:30:16…on March 7, 2014

17. Calendar time jumps
“System” time
“Calendar” time
…seconds since reference
…on March 30, 2014

18. Calendar time jumps
“System” time
312
“Calendar” time
10:59:58
GMT
…seconds since reference
…on March 30, 2014

19. Calendar time jumps
“System” time
312
“Calendar” time
10:59:58
GMT
313
10:59:59
GMT
…seconds since reference
…on March 30, 2014

20. Calendar time jumps
“System” time
314
312
“Calendar” time
10:59:58
GMT
313
10:59:59
GMT
…seconds since reference
…on March 30, 2014

21. Calendar time jumps
“System” time
314
312
“Calendar” time
10:59:58
GMT
313
10:59:59
GMT
…seconds since reference
Timezone changes
…on March 30, 2014

22. Calendar time jumps
“System” time
314
312
“Calendar” time
10:59:58
GMT
313
10:59:59
GMT
…seconds since reference
Timezone changes
03:00:00
Pacific Time
…on March 30, 2014

23. Calendar time jumps
“System” time
“Calendar” time
…seconds since reference
…on March 30, 2014

24. Calendar time jumps
“System” time
312
“Calendar” time
00:59:58
GMT
…seconds since reference
…on March 30, 2014

25. Calendar time jumps
“System” time
312
“Calendar” time
00:59:58
GMT
313
00:59:59
GMT
…seconds since reference
…on March 30, 2014

26. Calendar time jumps
“System” time
314
312
“Calendar” time
00:59:58
GMT
313
00:59:59
GMT
…seconds since reference
…on March 30, 2014

27. Calendar time jumps
“System” time
314
312
“Calendar” time
00:59:58
GMT
313
00:59:59
GMT
…seconds since reference
Daylight saving starts
…on March 30, 2014

28. Calendar time jumps
“System” time
314
312
“Calendar” time
00:59:58
GMT
313
00:59:59
GMT
…seconds since reference
Daylight saving starts
02:00:00
BST
…on March 30, 2014

29. Calendar time jumps
Managing calendar date and time is hard!
… but is not our problem!
… as long as we don’t use them!

30. System time jumps
“Absolute” time
“System” time
…ticks
…seconds

31. System time jumps
“Absolute” time
26 …ticks
“System” time
312
…seconds

32. System time jumps
“Absolute” time
26
“System” time
312
35 …ticks
313
…seconds

33. System time jumps
“Absolute” time
26
“System” time
312
35 44…ticks
313
…seconds

34. System time jumps
“Absolute” time
26
“System” time
35 44…ticks
Network date update
312
313
…seconds

35. System time jumps
“Absolute” time
35 44…ticks
213
26
“System” time
Network date update
312
313
…seconds

36. Solution:!
System time jumps
1. Do nothing
2. “Anchor” the time
3. Tell the user

37. “Anchored” Time
“Absolute” time
26 35 44…ticks
+295 +295 +195…offset
“System” time
Network date update
312 313 213
…seconds since reference

38. “Anchored” Time
“Absolute” time
26 35 44…ticks
17 18 19…seconds since boot
+295 +295 +195…offset
“System” time
Network date update
312 313 213
…seconds since reference

39. “Anchored” Time
Prepare:!
!
double initialOffset = SystemTime() - AbsoluteTime();
!
Start recording:!
!
// Wait for data!
!
New data received:!
!
double currentOffset = SystemTime() - AbsoluteTime();
double drift = currentOffset - initialOffset;
double anchoredTimestamp = timestamp - drift;

40. “Anchored” Time
On iOS:!
• Use CACurrentMediaTime() as AbsoluteTime().
• Use [NSDate timeIntervalSinceReferenceDate] as SystemTime().
!
On Android:!
• Use SystemClock.elapsedRealtimeNanos() as AbsoluteTime().
• Use System.currentTimeMillis() as SystemTime().
• Also see Location.getElapsedRealtimeNanos().

41. Distance calculation

42. Distance calculation

43. Distance calculation

44. Distance calculation

45. Distance calculation

46. Distance calculation

47. Distance calculation

48. Distance calculation

49. Distance calculation

50. Distance calculation

51. Distance calculation

52. Distance calculation

53. Distance calculation

54. Distance calculation

55. Distance calculation

56. Distance calculation

57. Distance calculation

58. Distance calculation

59. Distance calculation

60. Distance calculation

61. Distance calculation

62. Distance calculation

63. Distance calculation
Remove points that!
1. have unacceptably low accuracy
2. are too close to the points before or after them
Don’t remove points that!
1. the user has seen
2. are “too far”, unless you really know what you’re doing

64. Distance calculation
Location data point’s “accuracy”!
is not a scientific error value.

65. Distance calculation
The Earth is not flat!!
Don’t do cartesian maths.

66. Thank you!

68. Watch the video with slide synchronization on
InfoQ.com!
http://www.infoq.com/presentations/sensors-tracking-
user-activity