Final Year Project LYU0301 Location-Based Services Using GSM Cell Information over Symbian OS Mok Ming Fai CEG mfmok1@cse Lee Kwok Chau CEG leekc1@cse.

Final Year ProjectLYU0301

Location-Based Services Using GSM Cell Information over

Symbian OS

Mok Ming Fai CEG mfmok1@cseLee Kwok Chau CEG leekc1@cse

Agenda

Symbian OS Location-based services (LBS) Current GSM Positioning Methods Using GSM cell information in 2D space and

1D path MTR Travaller Future Work

The Symbian OS

Standard operating system for data-enabled mobile devices

32-bit, little-endian operating system working with ARM architecture chips with v4 instruction set or higher

Location-Based Services (LBS)

Services are provided based on user’s location under different wireless networks

LBS is applicable in various fields Different issues have to be considered Each of them requires different accuracy and

latency

GSM Positioning Methods

Region-based Cell Information (CI)

Point-based Time of Arrival (TOA) Angle of Arrival (AOA) Enhanced Observed Time Difference (E-OTD) Assisted GPS (A-GPS)

Point-based GSM Positioning Methods TOA (200m - 10km)

E-OTD (50m - 100m)

AOA (>>150m)

A-GPS (10m - 50m)

Motivation

Advanced positioning methods require: extra cost to existing network / synchronization between ba

se stations special hardware to end users telco-dependent

Not all LBSs need very accurate location information GSM cell information (e.g. cell ID) is available in ordi

nary GSM handset Symbian phone offers programming capability for ge

neral developers Location estimation by GSM cell ID is adopted in our

project

Overview of GSM Cell ID Location Estimation Each base station has unique location ID and cell ID Main idea: each base station can somehow provide

certain ‘information’ about a particular location Advantages:

simple implementation, only current registered cell is required

applicable on ordinary GSM phone without any support from telco

Location: [50]Cell ID: [1]

Location: [50]Cell ID: [2]

Location: [50]Cell ID: [3]

Location: [50]Cell ID: [4]

GSM Cell Change Event

Received signal strength from current registered cell is weaker than another, so cell change occurs

Consequences: More information provided More reliable in detecting boundary

Location-based Services in 2D Space

Initiatives To locate the approximate location of a mobile ph

one using a program that run on Symbian OS

Principle Determining GSM cells coverage and their distrib

ution Plot a cell ID-to-location map Locate current position of

a mobile device

Data Collection Method

Collected location ID and cell ID pairs for two telcos in the CU campus.

Data Collection method: Static Method for SmarTone Cell Change Method for Peoples

Principle of the Two Data Collection Methods

Static Method Wait for a sufficiently long period of time at a spec

ific point in the 2D map to see the strength and stability of a cell strength.

Determine the location ID and cell ID of that specific location after observing for a period of time

Principle of the Two Data Collection Methods

Cell Change Method Walk around the campus and find the “boundaries

” of different cells When cell change occurs we note down the chan

ge and try to find out the boundaries of the cells: location where cell change event is detected

Cell boundaries

Cell A

Cell B

Cell C

A->B

B->C

Advantages and Disadvantages of the Two Methods

Static Method Cell Change Method

Advantages Results very accurate at those selected points

Can figure out the boundaries of different cells in the area

Experiments only done on those selected points

Fast, no need to wait for a long time to get the result

Disadvantages Takes a longer time Boundaries detected are regions instead of sharp lines

Cannot figure out the distribution of cells clearly without dense selected points

Have to walk through the whole area several times

Experimental Results

For Peoples

Experimental Results

Experimental Results

For SmarTone

Experimental Results

Conclusion of the Experiment

Potential difficulties in 2D Space

ID-to-location map drawn not accurate enough Cannot locate the location of a mobile device to

an acceptable accuracy owing to the large size of cells

Hierarchy of cells make it even harder to locate our current position

Using GSM Cell IDs in 1D Space

Location: [50]Cell ID: [1]

Location: [50]Cell ID: [2]

Location: [50]Cell ID: [3]

Location: [50]Cell ID: [4]

Cell ID: [1->2] Cell ID: [2->3]

A set of multiple cell change events can indicate a path

Problem of Using GSM Cell IDs in 1D Space

The mapping of cell change event set and path is one-to-many

Apply this method on fixed path

MTR Traveller for Stations in Subway

Apply on traffic route MTR Traveller – detect station arrival Initial observation:

Between two stations in subway, there is exactly one cell change

This event can tell user that he / she is going from one station to another station

Station 2Station 1Cell ID changes here

MTR Traveller for Stations in Open Area

KCR stations in open area Many cells are involved in between two stations A station platform may also be covered by multiple cells

Group the cells into ‘station cells’ (pure cell ID) and ‘transition pairs’ (cell changes)

Station Cells:

[S1, O], [S1, B]

Transition Pairs:

[S1, S2, O, B], [S1, S2, B, P], [S1, S2, P, G]

Station 2Station 1

Operation of MTR Traveller

Transition pair => on the way between S1 and S2 Station cell => in the station platform

Station Cells:

[S1, O], [S1, B]

Transition Pairs:

[S1, S2, O, B], [S1, S2, B, P], [S1, S2, P, G]

Station 2Station 1

Cell ID: O

[S1, O] => in Station 1

Cell ID: OB[S1, B] => in

Station 1

Cell ID: BP[S1, S2, B, P] => on the

way of S1S2

Cell ID: PG[S1, S2, P, G] => on the

way of S1S2

MTR Cell ID Data

Peoples

MTR Cell ID Data

SmarTone

MTR Cell ID Data

Sunday

KCR Cell ID Data

Peoples

KCR Cell ID Data

SmarTone

Estimating the Accuracy of Proposed Method

Record the time difference at which the cell change occurs and at the moment that the train actually arrives the destination station

Convert the error range in time to distance by assuming constant velocity in that range

Result: 30m - 300m, comparable to E-OTD

Demonstration

Videos in actual stations

Evolution of Our Positioning Methods

Pure GSM Cell

Information Location

Estimation(Region Based)

GSM Cell Change Method

(Boundary / Line Based)

GSM Cell Change Method in

1D Path(Point Based)Detect registered

cell change occurred at cell boundary

Concentrate on specific cell changes (intersections between the path and the boundary)

Automatic Cell Data Collection

Collection of cell data was done manually in the past

Automatic cell data collection tool is required for regular update

Cell Snap

Contribution of Work

Enhancing pure cell ID location estimation by considering cell change events

MTR Traveller provides different application opportunities, such as: Notification Information providing

Cell Snap allows automatic cell data collection

Future Work

Improvement on MTR Traveller Personalization Informative User interface Distributed intelligence (SMS / GPRS)

Generic middleware / library for developers Other applications

Bus / tram route Detection of car speed detectors

Conclusion

GSM cell provides location-related information, but not accurate and reliable enough

Those information can be obtained through Symbian phone

The method was enhanced by using cell change events

Difficulties were encountered in 2D space The proposed method was also applied to 1D path:

MTR Traveller Automatic cell data collection by Cell Snap

Q&A Section

Thank you very much!