Master Thesis Presentation "Implementation and evaluation of Space Time Alarm Clock"

1. Implementation and
evaluation of
Space Time Alarm Clock
Master thesis presentation
Student: Adrian C. Prelipcean
Supervisor: Takeshi Shirabe
Co-supervisor: Falko Schmid
AG242X Geoinformatics

2. Outline
Implementation of Space Time
Alarm Clock
Evaluation of Space Time Alarm
Clock
Conclusions
Introduction
Objective
Methods - Space Time Alarm
Clock

3. Alarm clock
Why do we use alarm clocks?
1. To perform activities without worrying about the time
2. To get a sense (control) of time
3. To synchronize our schedule with that of others
4. As a reminder
5. To wake up

4. Navigation
Why do we use navigation systems?
1. To travel from the current location to a destination
2. To get directions while traveling
a. Audio directions
b. Visual directions - display the route (usually the shortest path) on a digital map

5. Alarm clock and navigation
Is it possible to combine the alarm clock functionality with the routing
functionality - reach a destination by a deadline?
source: http://linuxhub.net/wp-content/uploads/2010/01/alarm-clock.png and http://www.roadmapgps.com/models/tomtom-go-510/scr-navigation-map-

6. Objective
This thesis has three objectives:
1. develop a method that measures the time it takes from any location to a destination
by following any of the possible moves from a given location
2. provide the derived information to smartphones
3. develop a prototype that implements this method and test its computational
performance to identify bottlenecks

7. Methods
The thesis proposes a new method, called Space Time Alarm Clock, which
provides two main functionalities:
1. Alarm functionality - continuously tracks the user in space and time and alarms
when the user has to leave the current location to reach a destination by a deadline
2. Labeling functionality - determines the possible movement choices of a user,
computes the shortest travel time and informs the user about the consequences of
his/her movement via labels

8. Assumptions
Space Time Alarm Clock makes three important assumptions:
1. The user moves along streets, not through buildings or open field
2. The user moves at a constant speed
3. The user is a pedestrian

9. Space Time Alarm Clock - Steps
Space Time Alarm clock performs 6 computational steps:
1. Destination and deadline specification
2. Shortest travel time computation
3. Location detection
4. Earliest arrival time estimation
5. Consequence values computation
6. Communication

10. Alarm functionality
Space - Time prism

11. Alarm functionality
Space - Time cone for a destination

12. Alarm functionality
Space - Time cone for an origin and a destination

13. Shortest travel time computation
- Computed by using Dijkstra’s
Shortest Path Tree Algorithm
- each node contains the information
about the shortest travel time to the
destination

14. Location detection and map matching
- used to identify the road segment the user is on
- the locations received by the GPS are map
matched in real time
- one assumption is that the user moves along
streets, not through buildings or open field
source: http://graphics.stanford.edu/projects/lgl/papers/cdgnw-ammrfd-11/image.gif

15. Audio communication (upper left)
Visual communication (lower right)
Audio communication - alert when the user is
outside the space time cone
Visual communication - display
the shortest travel time for any of
the possible movement options

16. Implementation - architecture
General steps:
1. User input
2. Send parameters to the
server
3. Server replies with the
subnetwork
4. Labels are drawn and user
is alerted

17. Implementation - server
The server:
● stores the data set (Open Street Map) of the study area (Stockholm)
● extracts the relevant subnetwork
● performs the shortest path tree algorithm on the subnetwork
● sends the subnetwork, which also contains the shortest travel time from any node to
the destination, to the client

18. Implementation - client
The client:
● contacts the server for the subnetwork
● temporarily stores (caches) the subnetwork in a local database
● detects the user’s location along the subnetwork (map-matching)
● draws the labels (the shortest travel time for the possible movement choices the
user can make) at the decision making points
● alarms the user when he / she should start moving towards the destination to reach
it by the deadline

19. Implementation
User input flow

20. Using STAC
Where to next?
● Forward – 3 minutes and 6 seconds
to the destination
● Left – 6 minutes and 37 seconds to
the destination
● Turn around – almost 10 minutes to
the destination
Best action? The user decides

21. Using STAC
Available information
● Time left at the current location
● Shortest travel time to reach the
destination
● Shortest network travel distance to
reach the destination
What to do? The user decides

22. Evaluation
User evaluation:
● do users find STAC useful?
● insufficient respondents (5 questionnaires)
Performance evaluation:
● identify potential bottlenecks

23. Performance Evaluation
Processing time is influenced by:
● Available time
● Distance between origin and destination

24. Performance - overall time
Influence of available time (left)
Influence of distance between origin and destination (right)

25. Performance Evaluation
Operations performed by server:
● generate ellipse
● get closest node to the destination
● extract subnetwork within ellipse
● generate shortest path tree
Operations performed by client:
● read the subnetwork sent by the
server
● generate network topology
● build indexing system

26. Bottlenecks
Identifying bottlenecks (origin corresponds with destination and available
time is 3 hours)

27. Identified bottlenecks
Operations performed by server:
● generate ellipse
● find closest node to the destination
● extract subnetwork within ellipse
● generate shortest path tree (20.2%)
Operations performed by client:
● read the subnetwork sent by the
server (34.5%)
● generate network topology (24.4%)
● build indexing system (20.6%)

28. Conclusions and future work
● this thesis proposed, implemented and evaluated a new method, which we entitled
Space Time Alarm Clock (STAC), for monitoring a user’s location and alerting when
the user has to leave the current location to reach a specified destination by a
specified deadline
● this work identified the limitations and bottlenecks of the current implementation of
STAC
● to make STAC useful, future work should involve solving the current bottlenecks,
making STAC available for the entire world and testing whether STAC can be
extended to provide its functionality to users that are not only pedestrians

29. Summary
Is it possible to combine the alarm clock functionality with the routing
functionality - reach a destination by a deadline?
?
source: http://linuxhub.net/wp-content/uploads/2010/01/alarm-clock.png and http://www.roadmapgps.com/models/tomtom-go-510/scr-navigation-map-

30. Summary
The solved objectives:
1. develop a method that measures the time it takes from any location to a destination
by following any of the possible moves from a given location
2. provide the derived information to smartphones
3. develop a prototype that implements this method and test its computational
performance to identify bottlenecks

31. Summary
Is it possible to combine the alarm clock functionality with the routing
functionality - reach a destination by a deadline?
source: http://linuxhub.net/wp-content/uploads/2010/01/alarm-clock.png and http://www.roadmapgps.com/models/tomtom-go-510/scr-navigation-map-

32. Thank you for your time!
Q&A