Journal of Global Positioning Systems (2004)
Vol. 3, No. 1-2: 226-231
High Frequency Deflection Monitoring of Bridges by GPS
Gethin W Roberts, Emily Cosser, Xia olin Me n g , A lan Dodson
IESSG, The University of Nottingham, University Park, Nottingham, NG7 2RD, UK
e-mail: [e-mail] Tel: + 44 115 9513933; Fax: [phone]
Received: 15 Nov 2004 / Accepted: 3 Feb 2005
Abstract. The use of GPS for the deflection and
deformation monitoring of structures has been under
investigation for a number of years. Previous work has
shown that GPS not only measures the magnitude of the
deflection of the structure, but also it is able to measure
the frequency of the movement. Both sets of information
are useful for structural engineers when assessing the
condition of the structure as well as evaluating whether
Finite Element (FE) models of such structures are indeed
correct. GPS has the advantage of resulting in an abso lute
3-D position, with a very precise correspon ding time tag.
However, until recently, the maximum data rate was
typically 10-20 Hz, meaning that the maximum
detectable frequency was about 5-10 Hz. GPS also has
the disadvantage of multipath and cycle clips, and the
height component’s accuracy is typically 2 – 3 times
worse than that of plan. Previous work at the IESSG has
included the integration of RTK GPS, gathering data at a
rate of up to 10Hz, with that of data from an
accelerometer, typically gathering data at up to 200 Hz.
Accelerometers tend to drift over time, and can not detect
low vibration frequencies, but the acceleration data can
be double integrated resulting in changes in positions.
The integration of GPS and accelerometers can help to
overcome each others’ shortfalls. This paper investigates
the use of high rate carrier phase GPS receivers for
deflection monitoring of structures. Such receivers
include the Javad JNS100, capable of gathering data at up
to 100 Hz. Static trials have been conducted to
investigate the precision of such a receiver, as well as the
potential applications of such a high data rate. Trials
were carried out in a controlled environment and actual
bridge monitoring, and comparisons made with a Leica
SR510 receiver.
Key words: Bridge Deflection Monitoring, Data
Sampling Rate, RTK GPS, Data Processing.
1 Introduction
In 2001, The University of Nottingham was awarded a
three year grant from the UK’s Engineering and Physical
Sciences Research Council (EPSRC). The overall
objective of this project is the creation of a system
employing advanced computational tools coupled with
GPS and accelerometer sensors able to remotely monitor
the health of operational bridges without on-site
inspection. For the field measurements being able to
validate a compu tational model, such as a Finite Element
(FE) model, number and locations of sensors, sampling
rate and positioning accuracy are the indexes needed to
be considered. The first factor is normally determined by
the civil engineers, according to the size and d es ig n of th e
monitored structure. The second and third indexes are
also related to the size and design of the structure but
should be decided by the surveying engineers through
choosing appropriate surveying instruments. For a small
bridge of several tens of metres in length, the amplitude
and vibration frequency of the vertical movements can be
a couple of millimetres and tens of Hz. For a large
bridge, such as the Humber Bridge, the amplitude and
vibration frequency in the vertical direction can be of the
order of up to a meter and tenth of Hz, respectively
(Meng et al. 2003). To date the highest GPS data rate
used in experiments has been 10 and 20 Hz, which means
that only bridge dynamics of lower than 10 Hz could be
detected, taking other error budgets into account.
To overcome the abovementioned shortfall, integration of
GPS with triaxial accelerometers has been investigated in
a post-processing way. This approach can significantly
expand the valid measurable frequency to higher than 100
Hz. However, the absolute positioning fixes have to be
provided by the GPS solutions and there is lack of real-
time data transmission approach and relevant algorithm to
integrate data from two kinds of sensors.
Two JNS100 GPS OEM boards were recently purchased
from Javad Navigation Systems (Javad Navigation

Roberts et al: High Frequency Deflection Monitoring of Bridges by GPS 227
Systems, 2004), which are able to output raw data and
positions 100 times a second without interpolation
(Figure 1). These GPS receivers can be used to measure
bridge movements and also identify frequency dynamics
not higher than 50 Hz, due to Nyquist theorem.
Figure 1 The JNS100 OEM board GPS receiver
This paper investigates the use of these high rate
code/carrier phase GPS receivers for deflection
monitoring of structures. Zero baseline, short baseline
and kinematic trials have been conducted to assess the
precision of such a receiver, as well as the potential
applications of such a high data rate. These trials were
carried out in a controlled environment as well as for a
real bridge monitoring, and comparisons are made with
Leica SR510 single frequency GPS receivers gathering
data at a sampli ng ra t e o f 10 Hz.
2 Evaluation of receivers’ noise levels in static status:
zero baseline (ZBL) and short baseline (SBL) tests
The raw code and carrier phase data are output from the
receiver to a laptop and recorded using software called
PCView. The raw data is automatically converted to
Rinex format for post-processing. When the receiver
outputs data at 100 Hz there were data overrun problems
first on the serial port and then also on the USB port.
Due to this the data collected for this paper was only
recorded at a 50 Hz data rate, which is still fast enough to
measure much higher frequency structural dynamics than
has ever been possible with GPS before. The data
overrun problems are currently being investigated and
using these receivers at 100 Hz data rate will be the
subject of future papers.
The JNS100 receivers record code and carrier phase data,
only on the L1 frequency. Software for processing single
frequency data in the context of bridge monitoring has
been developed by the authors. This software, called
Kinpos, was used to process all the GPS data for this
paper. For more information about the software
development please see Cosser et al. (2004) or Cosser
(2004).
2.1 Zero baseline test
Two separate zero baseline trials were conducted on two
consecutive days with the JNS100 receivers used on the
first day and the Leica SR510 single frequency receivers
used on the next. The receivers recorded at the same
times on the two days, but offset by 4 minutes, so that
they would be recording data with the same satellite
geometry. On both days the two receivers of same type
were connected by a splitter to the same antenna, a Leica
AT503 choke ring antenna which was located on the roof
of the IESSG building. The aim was to compare the data
from the Leica receivers and JNS100 receivers under
similar observation cond itions. Th e Leica dual and single
frequency GPS receivers had been used during many
bridge trials in the past and their applicability to bridge
monitoring is known.
The JNS100 receivers were always set up to record at a
50 Hz data rate for all the trials outlined in this paper. In
the Kinpos software the data was then processed at a 50
Hz data rate and also resampled before processing to 10
Hz so that it could be directly compared to the Leica data.
The standard deviation of the JNS100 coordinates
appears greater for the 10 Hz data than for the 50 Hz data.
In each case the spread of the data is the same, but a
lower standard deviation is recorded for the 50 Hz data as
there are more sample points.
The processed coordinates in WGS84 are then converted
to those in a local coordinate system. The standard
deviations of the east, north and vertical components for
the Leica and JNS100 receivers can be seen in Table 1.
For a fairer comparison the Leica data are compared only
with the resampled 10 Hz JNS100 data. It can be seen
that the Leica data has a lower standard deviation in
every component when compared to the JNS 1 0 0 , w i th the
largest difference being seen in the vertical direction.
Figure 2 shows the time series of vertical coordinate error
of the Leica and JNS100 data at 10 Hz, while comparing
to the true coordinates. It is clear from this graph and
from Table 1 that the Leica receiver has a smaller spread
of coordinates in the vertical direction. This implies that
there is a better resolution of the carrier phase by the
Leica receivers. However, the noise levels caused by the
receivers are less than 1 cm for all three directions and
both types of GPS receivers, demonstrating the
appropriateness of these receivers for high precision
applications.
Table 1. Standard Deviation s o f JNS100 and Leica Receivers from ZBL
Standard Deviations (m)
EastNorth Vertical
JNS100 (50 Hz)0.00180.00230.0034
JNS100 (10 Hz)0.00190.00210.0041
Leica (10 Hz)0.00130.00170.0029

228 Journal of Global Positioning Systems
Vertical Coordinate Error
-0.015
-0.01
-0.005
0
0.005
0.01
0.015
220700.00220800.00220900.00 221000.00 221100.00221200.00 221300.00221400.00
Time (GPS Seconds)
Displacement (m)
JNS100 Leica
Figure 2. Time Series of Vertical Error for JNS100 and Leica Receivers
2.2 Short baseline tests
A short baseline test is a truer representation of survey
conditions and so the performance of the receivers in
practice can be better assessed. Atmospheric errors and
clocks are still mitigated, but multipath is now present in
the solution.
A short baseline trial was conducted on The University of
Nottingham campus during July 2004. Two AT503
antennas were positioned on two established points, the
coordinates of which were known from previous static
surveys. The two points were roughly 50 metres apart.
At each end of the baseline, a JNS100 receiver and a
Leica SR510 receiver were connected by a splitter to the
same antenna, meaning that the baselines measured by
each receiver combination were the same.
The baselines for this trial were processed in Kinpos and
the results can be seen in Table 2 and Figure 3. It can be
seen from Table 2 that once again the standard deviations
in all three components are lower for the Leica receivers,
the largest difference being in the east component, at
1.2mm, demonstrating slightly higher multipath in East-
West direction. Figure 3 shows the time series of vertical
coordinate error for the Leica receivers and the JNS
receivers at 10 Hz. The systematic bias of multipath is
now visible within the data and follows the same pattern
with slightly different amplitudes for b ot h r eceiver pairs.
From Figure 3, it can be found that to improve the
positioning precision, multipath need to be mitigated
either using appropriate mitigation algorithm or through
an internal filter of the receiver hardware and a choke
ring antenna. Dodson et al. (2001) investigated the use of
an adaptive filtering tech nique for reducing the impact of
multipath for str uctural deformation mon itoring.
Table 2. Standard Deviation s o f JNS100 and Leica Receivers from SBL
Standard Deviation (m)
EastNorth Vertical
JNS100 (50 Hz)0.00370.00560.0064
JNS100 (10 Hz)0.00370.00560.0067
Leica (10 Hz)0.00250.00500.0057
Vertical Coordinate Error
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
121300121400121500 121600 121700121800 121900 122000122100
Time (GPS Seconds)
Displacment (m)
JNS100 Leica
Figure 3. Time Series of Vertical Error for JNS100 and Leica Receivers
3. Evaluation of receivers’ noise level in dynamic
status: platform and bridge trials
3.1 Platform test
To test the potential of the JNS100 receivers in a dynamic
environment, a platform was set up on The University of
Nottingham campus (Figure 4). A wooden frame was
suspended from a tall tripod by means of a bungee cord,
which allowed free oscillation of the platform. The
reference receiver was located approximately 10 metres
away from the test rig, where an AT503 antenna was
connected via a splitter to the Leica SR510 and JNS100
receivers. An AT502 navigation antenna was mounted
on the test rig, which was then, via a splitter, conn ected to
the JNS100 and Leica SR510 receivers.
Using the test rig, two different trials were conducted.
For the first test, the platform was in rotation either held
still or disturbed from its resting position by someone
forcing the platform to move up and down. For the
second trial, the platform was just left to swing.
The first trial was conducted over a 10 minute time
period, where the bungee platform was held still for two
minutes and then made to oscillate for 2 minutes and so
on in rotation. The results for this trial for the JNS
receiver measuring at 50 Hz and resampled at 10 Hz, and
for the Leica receiver measuring at 10 Hz can be seen in
Figure 5. The amplitude of oscillation of the bungee
platform was measured as between 15 and 20 cm by both
GPS receivers. The JNS receiver has a period within the
last two minutes where there are a number of jumps
within the time series, which are caused by undetected
cycle slips. Apart from these jumps the measured

Roberts et al: High Frequency Deflection Monitoring of Bridges by GPS 229
displacement is very similar for both receivers. This
demonstrates the capability of the JNS receivers to
measure in a dynamic environment.
Figure 4 Platform for a D ynamic Test Using JNS100 and Leica
receivers
Vertical Displacement for the Bungee Trial
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
223100 223200 223300 223400 223500 223600 223700 223800
Time (GPS Seconds)
Displacement (m)
Leica JNS100
Figure 5 Time Ser ies of the Vertical Displacement by JNS100 an d Leica
receivers
In the second platform trial the bungee was just left to
swing with the wind. The results for this trial for the east,
north and vertical coordinates measured by the Leica and
JNS100 receivers can be seen in Table 3. In the trials, the
sampling rates for JNS100 and Leica receivers were set to
50 Hz and 10 Hz, respectively. For this trial the results
for both types of receiver match well, with the standard
deviations in the vertical and north component actually
being slightly better for the JNS100 receiver. Figure 6
shows that the multipath characteristics displayed by both
receiver solutions, in the vertical direction, are the same.
This is an encouraging result for the JNS receiver,
showing that in this dynamic environment they can
measure to the same degree of precision as survey grade
GPS receivers.
Table 3. Standard Deviat io n s o f JN S 10 0 an d Leica Receivers from
Platform Test
Standard Deviations (m)
EastNorth Height
JNS100 (50 Hz)0.00740.00780.0113
JNS100 (10 Hz)0.00740.00780.0115
Leica (10 Hz)0.00740.00790.0118
V
ertical Displacement for Bunge e Trial 2
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
222100222200222300 222400 222500222600 222700 222800222900
Time ( GP S S ec onds)
Displacement (m)
Leica JNS100
Figure 6 Time Series of the V ertical Movement by JNS100 and Leica
receivers
4 Bridge trial
A GPS and accelerometer bridge trial was conducted on
the Wilford Suspension Footbridge in Nottingham, over
two days in July 2004 (6th and 7th). This bridge has
been the focus of many trials conducted by The
University of Nottingham, due to its proximity and
relatively large amplitude movements. For more
information on previous trials conducted on the Wilford
Bridge, see for example Roberts et al. (2001) The
purpose of this trial was to analyse the performance of the
JNS100 receiv er s in a br idge en v ironm e n t .
In this trial, one reference station was set up on the bank
of the river, on a point whose coordinates were well
established from previous trials (Figure 7). The rover
receiver was located at the mid span of the bridge, where
the most movement is expected (Figure 8). At both
locations an AT503 antenna was connected via a splitter
to both the JNS100 and Leica single frequency receivers.
A number of sessions of data were collected on each day,
a selection of which will be analysed.
The GPS results for first session on the 7th July, which
was the second day of th e trial, can be seen in Table 4. I t
contains the standard deviations of the east, north and
vertical components for the JNS100 and Leica receivers.
For this particular session, the JNS100 receivers actually
performed better than the Leica in all three component
directions, the largest difference being seen in the north
direction. Both receivers were seeing exactly the same
satellites. The difference in standard deviations in the
vertical direction was very small as the same multipath

230 Journal of Global Positioning Systems
pattern can be seen in both times series (Figure 9). For
all the sessions during the bridge trial, the results from
the JNS100 and Leica receivers were very similar. In
some cases the JNS100 was slightly more accurate that
the Leica and in some cases this was the other way
around. The difference between the two receivers in all
cases was very small, showing that in the bridge
environment the performance of the JNS100 is
comparable with the Leica receivers, even at a much
higher sampling rate.
Figure 7. Reference st a t io n
Figure 8. Rover station
Table 4. Standard Deviat io n s o f JN S 10 0 an d Leica Receivers from
Bridge Trial
Standard Deviation
EastNorth Vertical
JNS100 (50 Hz)0.00250.00290.0043
JNS100 (10 Hz)0.00250.00290.0045
Leica (10 Hz)0.00270.00360.0046
Also, in the bridge trials the GPS results are compared to
a closely located triaxial accelerometer measuring at 50
Hz as well. The periods of the largest movement seen in
Figure 10 correspond to times in which people on the
bridge jumped up and down in unison ‘forcing’ the
bridge to move and then left to oscillate at its natural
frequency. In this graph the forced movement is apparent
in both the accelerometer and JNS100 data. When the
forced movement stops the accelerometer displays a
sinusoidal decay, which is movement at the bridge’s
natural frequency. This sinusoidal decay is not clear in
the GPS data as it is masked by the noise. However,
frequency analysis reveals that that this sinusoidal pattern
is still present in the GPS data even though it cannot be
discerned by the eye (Meng et al., 2004).
Vertical Displacement, Session bdg7_1a
-0.02
-0.015
-0.01
-0.005
0
0.00 5
0.01
0.01 5
0.02
294100 294200 294300 294400 294500 294600 294700 294800 294900
Time (GPS Seconds)
Displacement (m)
Leica JNS100
Figure 9. Vertical displacement by JNS100 Resampled to 10 Hz and the
Leica receivers
Vertical Displacement, Session bdg7_1a
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
294420.00 294430.00 294440.00 294450.00 294460.00294470.00
Time (G PS Secon ds)
Displacement (m)
JNS100 Accelerometer
Figure 10. JNS100 and accelerometer displacement in the vertical
direction both recorded at 50 Hz. The graph focuses on a time where
there was the largest movement on the bridge. The accelerometer data
is offset by 0.015m
4 Conclusion s
This paper has outlined the preliminary work conducted
with the JNS100 receivers. Zero baseline and static short
baseline trials have been conducted to assess the
precision of the receivers compared to known high
quality survey grade receivers (Leica system 500 single
frequency receivers). The results showed that the Leica
receivers performed slightly better than the JNS100 in the
static trials, but the difference was small. The JNS100
receivers do have a high precision carrier phase
observables.
Kinematic trials were performed on a bungee test rig and
also on a bridge. In a dynamic situation the JNS100
receivers performed as well as the Leica receivers. The
JNS100 results, measured at 50 Hz, were also compared

Roberts et al: High Frequency Deflection Monitoring of Bridges by GPS 231
to those from a closely located triaxial accelerometer
measuring at the same data rate.
JNS100 bridge trial results compared well to the
accelerometer findings, when identifying the periods of
largest movement. Most movement on the bridge was
masked by the GPS noise, but periods where large
displacements occurred could be discerned.
Acknowledgements
This project has been funded by the UK’s Engineering
and Physical Sciences Research Council (EPSRC) in
conjunction with Cranfield University (ref no
GR/R28218/01). The authors would like to thank Louise
Arrowsmith who took part in the field work whilst
working towards her MSc.
References
Cosser E (2004) Bridge Deflection Monitoring and
Frequency Identification with Single Frequency GPS
Receivers, ION GPS 2004, The 17th Technical Meeting of
the Satellite Division of the Institute of Navigation, 21-24
September, Long Beach, California.
Cosser E, Roberts G W, Meng X and Dodson A H (2004)
Single Frequency GPS for Bridge Deflection
Monitoring: Progress and Results, 1st FIG International
Symposium on Engineering Surveys for Construction
Works and Structural Engineering, 28 June-1 July,
Nottingham, England.
Dodson A H, Meng X and Roberts G W (2001) Adaptive
Methods for Multipath Mitigation and Its Applications
for Structural Deflection Monitoring, International
Symposium on Kinematic Systems in Geodesy, Geomatics
and Navigation (KIS 2001), 5-8 June, Banff, Canada.
Javad Navigation Systems (2004) OEM boards, [online].
Available at: <http://www.javad.com/> [Accessed 12
August 2004].
Langley R (1997) GPS Receiver System Noise, GPS World
8(6): 40-45.
Meng X (2002) Real-time Deformation Monitoring of
Bridges Using GPS/Accelerometers, PhD thesis, The
University of Nottingham, Nottingham, UK.
Meng X, Meo M, Roberts G W, Dodson A H, Cosser E, Iuliano
E and Morris A (2003) Validating GPS Based Bridge
Deformation Monitoring with Finite Element Model,
GNSS 2003, 22-25 April, Graz, Austria.
Meng X, Roberts G W, Dodson A H, Andreotti M, Cosser E
and Meo M (2004) Development of a Prototype Remote
Structural Health Monitoring System (RSHMS), 1st
FIG International Symposium on Engineering Surveys for
Construction Works and Structural Engineering, 28 June-1
July, Nottingham, England.
Roberts G W, Meng X and Dodson A H (2001) The Use of
Kinematic GPS and Triaxial Accelerometers to
Monitor the Deflections of Large Bridges, 10th
International Symposium on Deformation Measurement,
FIG, 19-22 March, California, USA.