Journal of Global Positioning Systems (2004)
Vol. 3, No. 1-2: 101-105
Alternative Positioning Method using GSM Signals
Goh Pong Chai
Nanyang Technological University, Singapore
e-mail: [e-mail] Tel: + 65-67905247; Fax: + 65-67910676
Received: 15 Nov 2004 / Accepted: 3 Feb 2005
Abstract. Location-Based-Services (LBS) have not made
any significant foray into commercial applications
following the mandatory 911 requirements for emergency
location initiated by US. Commercial adoption of LBS
has been hampered by cost issues and technical
difficulties faced by the various modes of positioning.
For LBS to be really usable, positioning, irrespective of
the technology, has to be a seamless and transparent
process and made available under most environments and
to within acceptable accuracy. Several services for
positioning technologies had been researched and
developed around GSM radio network, each with varying
success. This paper describes a commercial application
that has been developed on the Symbian platform. It
allows user-friendly access to maps on a phone and
incorporates a proprietary location algorithm that resides
within the phone. The algorithm is a significant
improvement from basic cell-id location methods.
Key words: LBS, GSM, Phone-Location, Cell ID,
Symbian
1 Introduction
Precise positioning technologies have seen developments
reached a plateau after Global Positioning System (GPS)
was developed and commercially deployed since the mid
70s. The main thrust for alternative positioning
techniques comes from the Federal Communications
Commission’s (FCC) wireless Enhanced 911 (E911)
rules mandating the effectiveness and reliability of
wireless 911 services by providing 911 dispatchers with
additional information on wireless 911 calls. The wireless
E911 program is divided into two parts - Phase I and
Phase II. Phase I requires carriers, upon appropriate
request by a local Public Safety Answering Point (PSAP),
to report the telephone number of a wireless 911 caller
and the location of the antenna that received the call.
Phase II requires wireless carriers to provide far more
precise location information, within 50 to 100 meters in
most cases (http://www.fcc.gov).
Various positioning technologies were subsequently
developed to ensure compliance, and/or to circumvent
technological limitations of each technology. Most GSM
operators settled for less accurate network positioning
solutions such as Cell-ID, EOTD and TDOA in order to
meet FCC E911 requirements. Unfortunately, most of
these attempts did not perform to standards that are
commercially viable. Trails conducted by a Singapore
telephone operator revealed a wide range of achievable
positional accuracy under varied conditions. The vagaries
of positions obtained render such technologies
unacceptable for many location based services (LBS).
According to a research study of 25 carriers in the
European Union carriers responded that they believe that
accuracy would increase the attractiveness of some
applications to consumers and revenue generation
(http://www.trueposition.com).
An improved positioning technology (codenamed gprX)
based on GSM technology, was developed by the
Nanyang Technological University (NTU) and agis, a
privately held company based in Singapore. A complete
model and processing algorithms were developed for the
whole of Singapore which allows a position to be
determined entirely on the processing capability of the
mobile handset. Efforts to developed passive systems that
do not need to register to the base station network such as
those using hyperbolic location positioning techniques
are typically done within controlled environments
(Senturk, 2002) (Mizusawa, 1996). Koets (2002) reported
mean location accuracy of 89 meters in the trials done in
San Antonio.
gprX is a passive system that does not require support
services from the telephone operator’s backend network.
Unlike most other GSM network solutions, the
technology takes away the burden of relying on
operator’s network services. Neither does it require

102 Journal of Global Positioning Systems
additional hardware beyond the basic mobile handset.
The position of a handset can be determined using gprX
to a much better accuracy than what is achievable today
under similar environments. Accuracy can be further
improved to an order of 30m under good network
conditions and using a terrain constrained solution. In a
worst case scenario, when operating in a single base
mode (only one base station is available), the returned
position is typically 60% better than otherwise achieved
by traditional cell-ID methods. With gprX, widespread
location based services (LBS) is now a real offering.
Current implementation requires Symbian OS or Pocket
Mobile OS (some hardware dependencies of the
PocketPC Mobile).
2 Review of Current Positioning Technologies
2.1 GPS-Phone
This solution uses a combined GSM+GPS handset, the
former provides the communications link while the latter
provides the positioning capability. This hybrid solution
fails in several areas:
Battery life of the device is too short.
It works only in open spaces with clear view of the sky.
Time to first fix (TTF) is not fast enough
2.2 A-GPS
This is another attempt to improve the coverage problem
and slow TTF encountered by the hybrid GSM+GPS
system by employing very sensitive GPS receivers that
may pick up GPS signals indoors (close to the windows
and open access). The weak signals are partially
overcome, or assisted, by augmenting the system with
relevant data that are transmitted to the hardware.
Assisted data includes information such as GPS
ephemeris and the approximate location. However, the
method still needs GPS signals, albeit weak. When GPS
signals are totally lost, A-GPS fails.
2.3 Terrestrial Based Network Solutions
The basic idea behind these technologies is that of
trilateration, using signals from ground based radio
stations such as GSM networks, TV signals and special
ground based transmitters to measure ranges to known
locations. In theory, if the time from a base station to a
receiver is known, then all that is needed is to know are
the station locations and the ranges to at least 3 of them in
order to compute the receiver’s location. Technologies in
this category include:
Phone Network Solutions: Enhanced Observed Time
Difference (EOTD), Time Difference of Arrival (TDOA),
Angle of Arrival (AOA) and Cell of Origin (COO) are
variants of such implementations. To compute a distance
based on time difference, the stations must be
synchronized. For EOTD, this requires supplementing the
network with additional hardware/services so that the
timings may be synchronized. For TDOA, each base
station is synchronized with a GPS receiver or atomic
clock.
Pseudolites: These are essentially ground-based
transmitters that produce unsynchronized GPS-like
signals. The technology uses a reference station for
generating corrections. Generally, these are expensive
and difficult to maintain. There are now attempts to
develop low-cost, synchronised pseudolites. If successful,
these could be a major breakthrough in positioning
technology.
None to these technologies today is able to provide a low-
cost, plug & play location solution that can be easily
deployed. Perhaps the closest today may be that of COO.
The advantage of COO is that it is pervasive, works in all
areas with signal coverage (including in-buildings). Two
main problems pertaining to this humble COO solution
are, firstly, it can be extremely inaccurate where the base
station spacing is sparse or where the signal propagated
by a base covers a large area and secondly, it requires
knowledge of the location of the base stations.
Deployment becomes problematic because the telephone
operator is in full control of the services and no access
can be granted without their permission. Accuracy using
COO is uncertain and can in the order of several
kilometres, rendering the service virtually useless.
3 gprX Technology
gprX positioning technology is a proprietary technology
that uses only information and signals that are available
on the phone to enhance the quality of COO. The world
today is wirelessly connected by two major technologies,
GSM and CDMA. The billionth GSM user was
connected in Q1 2004 – just a dozen years after the
commercial launches of the first GSM networks. Today,
GSM accounts for 73% of the world’s digital mobile
market and 72% of the world’s wireless market
(http://www.gsmworld.com).
GSM base stations are strategically located so as to
provide sufficient signal overlaps by neighbouring
stations to ensure smooth handovers to the next cell when
a mobile handset moves from one point to another and
also to provide for optimal voice quality. When a mobile
unit is switched on, it is automatically registered to a

Goh: Alternative Positioning Method using GSM Signals 103
particular network and maintains a list of cells based on
signal parameters and handover protocols. Unfortunately,
signal propagation and signal pattern at any location is
dynamic, and can be quite erratic, depending on ambient
conditions. By studying the characteristic behaviour of
signals at a location (affected by the ambience and
terrain), a model was developed and appropriate
algorithms were developed to map the observable
parameters to probable locations. When a handset is in
motion, characteristic patterns are generated and this can
be assigned to the geographical location computed by the
algorithm.
3.1 gprX Implementation
This process is currently implemented as a client software
resident in the handset. A thin client solution may also be
offered so that most of the processing is taken off the
mobile handset and moved to the server. Most high end
phones available today are equipped with sufficient
processing power to run the software locally. The client
software continually monitors the network parameters
and other information provided by the GSM network and
computes the most probable location in real-time. Once
the model has been calibrated, it does not require
knowledge of where the base stations are located and is
therefore totally independent of the network operator’s
network services. The dynamic nature of the GSM signals
is continually monitored and re-calibrated. Re-calibrated
model parameters are returned to the handset which will
improve its location capability with time.
3.2 Constrained Solutions
In situations where physical constraints are known, then it
is possible to locate with almost 100% certainty where
the handset is located. Buildings which are served by
only one single base station (and repeaters) fall into this
category. For improved performance, additional
geometric and/or terrain constraints may also be imposed
to improve the positioning accuracy. With known
constraints, the computed location can be dramatically
improved. Underground roads and train tunnels are
geometric constraints that may be used to improve
positioning quality.
3.3 Poor GSM Network
In the worst case scenario when only one base station is
within monitoring range, the algorithm cannot return an
improved position, then the COO weighted position will
be returned. This is not the location of the base station but
the location of a computed and averaged high sensitivity
point. In all cases, it is a much better solution than the
existing COO provided by the operator’s network
services which will only indicate the base-station
location. Current tests showed that an improved COO
gprX solution is at least 60% better than traditional COO
location (Figure 1).
3 Cell Sector
Coverage
Station1
Station2
Station3
Computed
imp r oved
position
Ba se
Station
Fig1 Improved COO Location
3.4 Transient Signal Errors
In areas of low population density, it is usual to have
long-range GSM stations and repeaters which are located
far from its actual location. In such situations, the COO
technique returns locations with errors of a few
kilometres. In Singapore for example, a handset that at
Orchard road may be tracked by a base located at Bukit
Timah, a good 6 kilometres away. A filtering algorithm
embedded in gprX removes such transient errors.
3.5 gprX Pre-Release Processes
gprX services are released after performing a sequence of
pre-release analysis and formulation of algorithm and a
model suitable for the particular terrain. Depending on
requirements and geographical coverage, the pre-release
exercise typically takes about a month to complete
(Figure 2a). This is sufficient to for a model to be
developed and to start a gprX session. A subsequent post-
release automated process will continue to re-model the
parameters continuously so that positioning accuracy will
improve with time.
3.6 gprX Post-Release (run-time) Processes
After gprX services are released and implemented, a self-
calibration process is implemented to improve the initial
modelling parameters and also to provide a feedback loop
for changes in the network. For example, when a new
base station is added or the terrain is changed drastically
because of new high-rise buildings, this self-enhancement
process will capture these changes and feedback into the

104 Journal of Global Positioning Systems
model. Typically, the computed results will improve with
time as this post-release phase kicks in (Figure 2b).
Fig 2a Pre-Release Process
Fig 2b Post Release Process
3.7 Test Results
Using a calibrated model for Singapore, a drive test was
done using a GPS to track an accurate position while the
gprX provides location from only the phone. Figure 3
shows a comparison of a gprX location track (red circles)
superimposed on a GPS track (blue points). In all tests,
similar excellent results were obtained.
Fig 3 A Test Result
4. Conclusion
Hitherto, no single tracking technology has been able to
fulfil the quest to determine accurate positions without
the use of complicated and expensive hardware and under
operating uner most environments. The FCC’s mandate
has not really been fulfilled even after massive
development efforts to find the holy grail of positioning.
While GPS is able to determine very good location fixes,
it suffers from its inability to work under cover, relatively
high costs and being hardware dependent. Other
technologies have proven to be limited in terms of its
overall performance in price and accuracy. There are
many advantages of gprX:
Telco/Network Independent. There is no longer a need to
engage the telephone operators and depend on expensive
backend services.
No Additional Hardware.
Pervasive Coverage. Works anywhere, across countries
where there are GSM signals, including in-buildings.
Good Accuracy in most situations. Sub-100m accuracy is
achievable in most environments. High accuracy location
data increases the value of mobile location services,
enhancing the user experience and maximizing revenue.
Scalable and Cost Effective. Start-up costs are necessary
but pales in comparison with other current technology
available.
Seamless, Client Server Integration. Interfaces for system
access and data distribution provide mechanism for third
party software to be built upon the basic gprX services.
Scalable. Not limited by network infrastructure.
Acknowledgements
gprX was developed jointly by the author and Agis
Private Limited (http://www.asiagis.com.sg), a Singapore
domiciled company. Agis has extensive geographic
database of Singapore and Malaysia and was instrumental
in the development of the model and software
development of services on mobile devices.
References
Mizusawa GA (1996): Performance of Hyperbolic Position
Location Techniques for Code Division Multiple Access.
Virginia Polytechnic Institute and State University. PhD
Thesis.
Senturk H (2002): Performance Evaluation of Hyperbolic
Position Location Technique in Cellular Wireless
Networks. Air Force Institute of Technology Wright-
Patterson AFB.

Goh: Alternative Positioning Method using GSM Signals 105
Koets MA (2002): Development of Inverse Hyperbolic
Positioning Using the GSM Cellular Telephone Network.
Southwest Research Institute, 16-9222.
Reference web-sites:
http://www.trueposition.com: (2004) U.S. Poised to Capitalize
on Location Services, White Paper
http://www.fcc.gov
http://www.gsmworld.com