Footprint-oriented generation of traffic trajectories using cellular - - PowerPoint PPT Presentation

Institute for Engineering Geodesy

Footprint-oriented generation of traffic trajectories using cellular phone data trajectories using cellular phone data

Geneva, 5–9th May, Geospatial World Forum 2014

Shenghua Chen

Institute of Engineering Geodesy University of Stuttgart, Germany

Content

Institute for Engineering Geodesy

Content

Motivation & problems

1

Relationship between data structure

2

Generation of traffic trajectory and test

3

Map matching for validation

4 5

Conclusion

2 Footprint-oriented generation of traffic trajectories using cellular phone data

Motivation and problems

1

Institute for Engineering Geodesy

Motivation and problems

1

Figure: Cellular probes used in traffic condition acquisition

Rich source: High penetration of cellular phone; Highly combined with market: LBS has grow quickly recently;

Figure: Cellular probes used in traffic condition acquisition

3 Footprint-oriented generation of traffic trajectories using cellular phone data

Highly combined with market: LBS has grow quickly recently; Cost: Few additional cost to system;

Background for problems

1

Institute for Engineering Geodesy

Background for problems

1

Handover operation don’t happens on the boundary completely; The cell sequence can reflect the trajectory roughly;

4 Footprint-oriented generation of traffic trajectories using cellular phone data

The cell sequence can reflect the trajectory roughly; The cell sequence by time slot may do not reflect the trajectory comprehensively, especially when dealing with A data with long time intervals.

Relationship between data structure

2

Institute for Engineering Geodesy

Relationship between data structure

2

Static data Static data

Digital Road Map (DRM) Cellular Phone Network Best server plots Signal strength maps Antenna positions and orientation Antenna positions and orientation

Dynamic data

• Cellular Phone Network
• A interface
• A-bis interface

5 Footprint-oriented generation of traffic trajectories using cellular phone data

• A-bis interface

Relationship between data structure

2

Institute for Engineering Geodesy

Relationship between data structure

2

Footprints produced by spatial overlapping between road network and GSM cells between road network and GSM cells Geospatial overlapping between the cellular Figure: Simplified system architecture of a mobile phone network with the interface (Do-iT, 2008)

6 Footprint-oriented generation of traffic trajectories using cellular phone data

Geospatial overlapping between the cellular phone network and traffic road network is the principle behind this ideal of footprint-oriented generation of trajectory. (Do-iT, 2008)

Data structure of traffic telematic database

2

Institute for Engineering Geodesy

Data structure of traffic telematic database

2

Data structure in database for traffic information acquisition based on mobile phone data based on mobile phone data

7 Footprint-oriented generation of traffic trajectories using cellular phone data

Footprint-oriented generation of traffic trajectories

3

Institute for Engineering Geodesy

Footprint-oriented generation of traffic trajectories

3

8 Footprint-oriented generation of traffic trajectories using cellular phone data

Flow chart of footprint-oriented generation of trajectory

Mobile phone data processing

3

Institute for Engineering Geodesy

Mobile phone data processing

3 Abnormal handover Behavior: No common No common point in the boundary between between continuous records

9 Footprint-oriented generation of traffic trajectories using cellular phone data

Abnormal handover line number = [51 52 53 62 63 64 65 69 70]

Processing A interface data from GSM system

3

Institute for Engineering Geodesy

Processing A interface data from GSM system

3

Abnormal handover Behavior: No common point in the boundary between continuous records Reason: overflow or jump; Overflow: sequential Overflow: sequential Line: 7, 8, 25, 26, 63, 65 Continue; Jump: not sequential, handle the line, and continue Jump: not sequential, handle the line, and continue Line: 33, 40, 41 To get the polygon which have common point with these cells. Consider the last abnormal HO in the A data

10 Footprint-oriented generation of traffic trajectories using cellular phone data

Footprint-oriented generation of traffic trajectories

3

Institute for Engineering Geodesy

Footprint-oriented generation of traffic trajectories

3

For problem two

11 Footprint-oriented generation of traffic trajectories using cellular phone data

Result of footprint-oriented trajectories(1)

3

Institute for Engineering Geodesy

Result of footprint-oriented trajectories(1)

3

Figure: visualization of original cells (data 1: normal) Figure: visualization of original cells (data 2: long time interval)

12 Footprint-oriented generation of traffic trajectories using cellular phone data

normal) time interval)

Result of footprint-oriented trajectories(2)

3

Institute for Engineering Geodesy

Result of footprint-oriented trajectories(2)

3

Figure: Cell sequence from data 2 by

• rdinal analysis

13 Footprint-oriented generation of traffic trajectories using cellular phone data

Cells in A data and the cell sequence by TSP algorithm

Result of footprint-oriented trajectories(3)

3

Institute for Engineering Geodesy

Result of footprint-oriented trajectories(3)

3

R = 0.7*diagonal of cell MBX 14 Footprint-oriented generation of traffic trajectories using cellular phone data

Figure: Flow chart of extending the cells to connect cells along the cell sequence

Result of footprint-oriented trajectories(5)

3

Institute for Engineering Geodesy

Result of footprint-oriented trajectories(5)

3

Figure: Parameter:

15 Footprint-oriented generation of traffic trajectories using cellular phone data

Figure: Visualization of footprint –

• riented trajectory by data1

Parameter: total distance: 46.2km Time peroid: 1h Maximum time interval: 0.3h Mean time interval: 0.1h

Result of footprint-oriented trajectories(4)

4

Institute for Engineering Geodesy

Result of footprint-oriented trajectories(4)

4

Figure: Visualization of footprint –oriented trajectory by data 2

16 Footprint-oriented generation of traffic trajectories using cellular phone data

Figure: Visualization of footprint –oriented trajectory by data 2 Minimal cells set which connect from the start to the end

Result of footprint-oriented trajectories(6)

3

Institute for Engineering Geodesy

Result of footprint-oriented trajectories(6)

3

Figure: Footprint-oriented generation of traffic trajectory based

• n A data from wireless communication system

Total distance: > 120km Time period: > 6h (7:01:22 – 12:56:46)

17 Footprint-oriented generation of traffic trajectories using cellular phone data

Parameters: Total distance: > 120km Time period: > 6h (7:01:22 – 12:56:46) Area: Stuttgart – Karlsruhe

• Manheim

Maximum time interval: 1.17h Mean time interval: 0.17h

4 Map matching for validation

Institute for Engineering Geodesy

Purpose: GPS data can not decide whether the trajectory by footprint are

4 Map matching for validation

Map matching – Purpose

correct.

Map matching – flow chart

Figure: Flow chart of map matching to validate the trajectory by mobile phone data Figure: Flow chart to restructure the

18 Footprint-oriented generation of traffic trajectories using cellular phone data

Figure: Flow chart to restructure the road network using quadtree grid Two times to narrow the search space: Quadtree grid connected to line; Quadtree grid only in the area of GPS point (temporary);

4 Map matching for validation

Institute for Engineering Geodesy

4 Map matching for validation

Map matching (1)

Purpose of map matching Steps of map matching Figure 10: gps and footprint-oriented trajectory

Parameters:

19 Footprint-oriented generation of traffic trajectories using cellular phone data

Parameters:

Size of quadtree grid: 300m * 300m Mean of line length: 40.17m Mean GPS intervals distance: 17.6m (0.5s) Size of quadtree grid: 300m * 300m

4 Map matching for validation

Institute for Engineering Geodesy

4 Map matching for validation

Map matching result

20 Footprint-oriented generation of traffic trajectories using cellular phone data

Figure: Map matching result with GPS, and trajectory generated Figure: Some limitation for distance map matching

4 Map matching for validation

Institute for Engineering Geodesy

4 Map matching for validation

Distance map matching to topological map matching

21 Footprint-oriented generation of traffic trajectories using cellular phone data

Figure: topological Map matching and illustration

4 Map matching for validation

Institute for Engineering Geodesy

4 Map matching for validation

Validation of footprint-oriented trajectory by map matching result Formula: Formula: Correction rate FT: set of lines which combine the trajectory generated by footprint- FT: set of lines which combine the trajectory generated by footprint-

• riented algorithm

MM: set of lines which combine the trajectory generated by topological map matching algorithm Result: Type of A data Normal handover record handover records with long time intervals time intervals Correction rate 93.8% 97.1%

22 Footprint-oriented generation of traffic trajectories using cellular phone data

5 Conclusion

Institute for Engineering Geodesy

5 Conclusion

A data (handover operation records for mobile phone) from GSM can be used to generate reliable and robust results, with assistance of digital map and map-aiding technology. digital map and map-aiding technology. Role of footprint derived from A data, connect the static and dynamic data for generation of traffic trajectory. data for generation of traffic trajectory. The rate of the correct street selected from network can be 94.8%, even get a correction rate of 90.1% for A data with long time intervals. The footprint-oriented trajectory generation can endure more than 1h time intervals, this is sparse sampling. Propose reason may be the limited trajectory. limited trajectory. Correction rate for trajectory of long trip is higher than that of short trip.

23 Footprint-oriented generation of traffic trajectories using cellular phone data

Institute for Engineering Geodesy

Contact: Shenghua Chen Shenghua Chen Institute of Engineering Geodesy, University of Stuttgart Geschwister-Scholl-Str. 24/D, 70174 Stuttgart, Germany Tel.: +49 (0)711 685-84053 Fax: +49 (0)711 685-84044 Email: shenghua.chen@ingeo.uni-stuttgart.de