Performance Comparison of Differential GNSS,

EGNOS and SDCM in Different User Scenarios in

Finland
Simo Marila, Mohammad Zahidul H. Bhuiyan*, Jaakko Kuokkanen, Hannu Koivula, Heidi Kuusniemi*
Finnish Geospatial Research Institute
Department of Geodesy and Geodynamics
*Department of Navigation and Positioning
Kirkkonummi, Finland
{simo.marila, zahidul.bhuiyan, jaakko.kuokkanen, hannu.koivula, heidi.kuusniemi}@nls.fi

Abstract—Positioning errors of stand-alone GNSS can be EGNOS performance is monitored at RIMS (Ranging and
reduced with different real-time augmentation approaches. For many Integrity Monitoring Stations) stations and results are also
navigation and positioning applications, Satellite-Based publicly visible at European Satellite Services Provider’s
Augmentation System (SBAS) and Differential code-based GNSS (ESSP’s) web pages [2]. The EGNOS performance analysis at
(DGNSS) can offer substantial improvement in the positioning RIMS stations are very optimistic, mainly for two reasons: i)
accuracy. Finland is covered by EGNOS (European Geostationary these stations’ GNSS pseudorange data are utilized to compute
Overlay Service) even though it is lying at the north-east margin of the corrections, and ii) the stations are stationary and located
the coverage area. In addition to EGNOS, the neighboring SBAS
usually outdoor with very good sky visibility. Therefore, the
system SDCM (System for Differential Corrections and Monitoring)
EGNOS accuracy reported in ESSP may not always reflect the
has monitoring stations near Finland. In this work, the researchers at
the Finnish Geospatial Research Institute (FGI) conducted some field
real picture of the expected accuracy within its area of
experiments with mentioned SBASs and National Land Survey of coverage. In recent years, researchers in Norway have been
Finland’s (NLS) DGNSS service in order to find out reachable monitoring EGNOS performance [5], [6] with the PEGASUS
accuracies and rough estimates of availability in varying positioning software [7]. It was shown in [5] that the EGNOS performance
conditions. Interest to SDCM grew due to its vicinity from Finland can be degraded adversely by ionospheric disturbances at high
and also partly due to its ability to provide GLONASS corrections latitudes. In addition, the geometry of GPS satellites can
which EGNOS has not yet been able to offer. sometimes be poor at northern latitudes, contributing to
degraded positioning performance. Recently, Finland has also
Keywords—SBAS, EGNOS, DGNSS, SDCM, positioning undertaken a similar project entitled ‘Finland’s EGNOS
accuracy, GPS, GLONASS, availability Monitoring and Performance Evaluation (FEGNOS)’ in order
to monitor the performance of EGNOS in Finland [8].
I. INTRODUCTION In this work, the researchers at Finnish Geospatial
Standard Positioning Services (SPS) of GNSS (receiver in Research Institute (FGI) conducted some field experiments
stand-alone mode) can be augmented with code-based with EGNOS, SDCM and National Land Survey of Finland’s
Differential GNSS (DGNSS) or with Satellite-Based (NLS) DGNSS service in order to find out reachable
Augmentation System (SBAS). In general, augmentation accuracies and rough estimates of availability in varying
system offers higher positioning accuracy than positioning positioning conditions. At first, a brief description about
made in stand-Alone mode. In addition, SBAS offers different GNSS augmentation services is presented in Section
information on availability and integrity of corrections and II. In the next section, equipment setup and coordinate
SPS ensuring positioning in safety critical applications. transformation related details are presented. In Section IV,
This paper tries to clarify the achievable real-time experimental details are illustrated, followed by results
accuracy in different user scenarios with the state-of-the-art analysis and interpretation in Section V. Finally, some
geodetic GNSS receiver in Finland with corrections from concluding remarks are made in Section VI based on the
Finnish National Land Survey’s (NLS) Differential GNSS findings of the experiments.
(DGNSS) Service, from European EGNOS (European
Geostationary Overlay Service) [1], [2] service and from
Russian SDCM (System for Differential Corrections and II. AUGMENTATION SERVICES
Monitoring) service [3], [4]. Accuracies are determined based
on the data obtained from a series of test measurements. Tests A. NLS’s DGNSS Service
are performed in both static and dynamic modes. In the beginning of 2014, FGI opened a positioning service
based on the Differential GNSS (DGNSS) corrections for GPS
and GLONASS [9]. The positioning service is based on real

978-1-4799-8915-7/16/$31.00 ©2016 IEEE

time data streams from permanent GNSS stations covering the used Network RTK (GPS+GLONASS) corrections to get
whole Finland, called the FinnRef [10], [11]. FinnRef network accurate reference solutions (fixed only used) for comparison
has 20 permanent GNSS reference station as shown in Fig. 1. of DGNSS, EGNOS and SDCM accuracies.
The positioning software utilized in the service is called
GNSMART and was developed by the company named B. European Geostationary Overlay Service (EGNOS)
GEO++ GmbH [12]. The primary function of FinnRef is to EGNOS is an augmentation system providing correction
offer geodetic-grade GPS/GNSS measurements used for data and integrity information for improving GPS (GPS L1
forming and maintaining the national coordinate system Coarse/Acquisition civilian signal) Positioning, Navigation
(EUREF-FIN). In addition, the FinnRef network is used for and Timing (PVT) over Europe. Main services of EGNOS are
many GNSS-related research activities. For example, it is now Open Service (OS) [1] and Safety of Life service (SoL) [14].
possible to analyze the positioning performance of different The SoL service has strict minimum performance parameters
augmentation services via FinnRef network. FinnRef network especially designed to meet the requirements of civil aviation.
has been established in mid 1990’s and it has recently been SoL service has been operational since 2011. EGNOS
upgraded during 2012-14. corrections refer to EGNOS Terrestrial Reference Frame
(ETRF), which is closely aligned to the International
Terrestrial Reference Frame (ITRF) and to WGS84 coordinate
system used by GPS. EGNOS comprises of two main
segments in space and ground. RIMS stations all over the
service area collect raw GPS data measurements on which
corrections and integrity data formation in the processing
centers are based on. The corrections and integrity information
are sent to users via geostationary satellites, namely PRN 120
and PRN 136, as of February 2016 [2]. Elevations of these
satellites are relatively low in Finland as compared to most
other European countries.

C. System for Differential Corrections and Monitoring

(SDCM)
System for Differential Corrections and Monitoring,
abbreviated SDCM (СДКМ, Российская система
дифференциальной коррекции и мониторинга) is Russian
operated SBAS system [15]. SDCM has a similar working
principle like other existing SBAS systems. Monitoring
stations are mostly located in the territory of the Russian
federation. The nearest ones to Finland are in Saint Petersburg,
Svetloe observatory and Lovozero in North. SDCM data is
transmitted to user via three geostationary satellites with PRN
numbers 125, 140 and 141. SDCM provides services for both
GPS and GLONASS. GPS and GLONASS use different
coordinate systems WGS84 and PZ-90.02, respectively.
SDCM generates corrections in WGS-84 by matrix
transformation of GLONASS data from PZ-90.02. In SBAS
messages, SDCM data for GLONASS and GPS are presented
Fig. 1. Permanent GNSS-stations (FinnRef), EGNOS RIMS stations and in single time scale, i.e., GPS time scale [15].
SDCM reference stations around Finland.

The user of the open positioning service can either choose III. EQUIPMENT AND COORDINATES
to use correction from a single station (nearest or any selected)
or correction interpolated from multiple station data to certain A. Equipment Setup
spot nearby the user (rover). Corrections enhance the L1 code- Experiments were performed with Javad Delta receivers
based positioning solution and they include differential (Delta-G3T), which were connected via signal splitters to one
pseudorange corrections both for GPS and GLONASS single antenna. The antennas used in the tests were NovAtel
satellites seen by the reference station(s). The DGNSS service GPS-702-GG and Javad JPL designed choke ring antenna with
sends this information in an RTCM 2 message, specifically ‘Dorne Margolin’ elements. During the static tests, the
types 1, 3 and 31. Corrections are provided to users via NovAtel antenna was mounted on a tripod, and on a specific
internet using the NTRIP-protocol [13]. metal plate in Masala rooftop at the FGI premises. At all
The GNSMART software is also capable of providing FinnRef stations, Javad JPL antennas are mounted on top of
different types of RTK corrections. These corrections are, at three meter steel masts. During all kinematic tests, NovAtel
the moment, generated only for research purposes and are not antenna was mounted on the roof-rack of a passenger car.
provided in the public positioning service. In this work, we High quality coaxial antenna cables were used to connect the
antenna with the receiver. Furthermore, we also needed a Metsähovi, visibility was good. Photos from some of the test
computer with internet connection to feed DGNSS and sites are shown in Fig. 3.
Network RTK corrections in real-time. We used GNSSSurfer
(V1.08) to send corrections to the receiver. Corrected Static accuracy of EGNOS and SDCM were tested in
coordinates and other information were written to a file in Masala rooftop and in forest conditions. In addition,
Javad’s proprietary text message format. 10° satellite cut-off simultaneous tests at two permanent FinnRef stations were
angle was used in all the tests. Dynamic positioning mode was conducted. In these tests, two receivers were left in station
configured in the receiver while doing kinematic tests and TUO2 (at Tuorla) and other two in VIR2 (at Virolahti). The
static mode was configured for static measurements. idea behind this was to find out if there was any reference
Positioning interval was set to 1 second. Maximum allowed station distance dependency in EGNOS or SDCM. TUO2 lies
correction age (DGNSS, EGNOS and SDCM) was set to 30 in the middle of LAP and GVL RIMS stations (280 and 290
seconds for all the tests. km) and VIR2 just beside the LAP RIMS station at Virolahti.
For SDCM, test sites were located respectively about 400 kms
and 120 kms outside from the nearest monitoring station
B. Coordinate Reference Frames (Svetloe). Receivers were connected via splitters to antennas
The coordinates were obtained in the form of geographical of permanent FinnRef stations.
coordinates (Latitude, Longitude and Ellipsoidal Height).
Network-RTK reference coordinates (in case of kinematic
tests) and DGNSS coordinates were obtained in EUREF-FIN,
but EGNOS and SDCM solutions were obtained in WGS84
coordinates. Geographical EUREF-FIN coordinates were
converted to planar ETRS-TM35FIN coordinates with official
formulas. Geographical WGS84 coordinates were converted
with same formulas to planar coordinates, but due to
difference of EUREF-FIN and WGS84 we also considered
offset between these systems by adding certain values for
North, East and Up. Offset-values were calculated from the
known differences between WGS84 and EUREF-FIN at all
FinnRef stations. For each test point or Test Drive (TD), the
nearest station offsets were directly applied. Uncertainty of
final transformed coordinates was assessed to be within some
centimeters.

IV. EXPERIMENTS

A. Static Test Scenarios

The main idea behind the static experiments was to find
out with relatively long time series what kind of accuracy can
be reached in good and challenging positioning environments
and also how location of the rover station with respect to the
reference stations affects the offered solution (distance to the
reference station(s)) in case of DGNSS. Three full days of data
were collected at each test session. In addition to DGNSS,
EGNOS and SDCM also stand-alone tests were performed in
order to understand the offered positioning accuracy of the Fig. 2. Static test points.
receiver without applying any correction. For static tests,
reference coordinates were measured with long PRS (Pseudo
Reference Station) measurements and for Masala metal plate
point, we had earlier measured the true coordinates with static
relative GPS. Reference point in thick forest was ensured with
tachymeter.
In DGNSS tests, both nearest station and network-based
approach were tested. Static DGNSS tests were done at three
locations as shown in Fig. 2.. Three test locations are in: i)
Lahti in the middle of four FinnRef stations (nearest 101 km),
ii) in Masala (nearest 10 km), and iii) in Metsähovi just beside
the station. In Masala, tests were done in two reference points,
at good visibility condition at the rooftop of FGI main
building and just behind it in the thick forest. In both Lahti and Fig. 3. Pictures from test sites in Masala (Forest), in Lahti and in FinnRef
station VIR2.
B. Kinematic
During the kinematic tests, three receivers were
simultaneously used. One receiver was used for collecting
reference coordinates (Network RTK). Other two receivers
were used to collect positioning solutions with corrections
from either Network DGNSS and EGNOS or EGNOS and
SDCM respectively. Only those time epochs with fixed PRS
solutions were used as reference. Due to this, at challenging
GNSS condition, we had time-to-time measurement gaps in
the reference position solution. Kinematic tests were
conducted with a passenger car in three different scenarios: i)
at open area in Siuntio, Finland (TD 1), ii) at Helsinki city
center, Finland (TD 2), and iii) at typical Finnish road Fig. 5. Antenna on roofrack in Siuntio (TD 1).
environment (TD 3). In all the above cases, positioning
solutions were achieved with Network DGNSS and EGNOS. V. RESULTS AND ANALYSIS
Comparison of EGNOS and SDCM was performed in a The results are presented in this section for the test
separate car drive test in eastern Finland (TD 4) at condition environments described in the previous section. The initial
typical for Finnish road environment. Static data when the car accuracy of the receivers used was evaluated first without
was stopped was not used in the accuracy calculation. The test applying any corrections. The stand-alone accuracy statistics
drive locations are shown in Fig. 4. are presented in Table I and the related horizontal scatter plot
in Fig. 6.
TABLE I. STAND-ALONE GPS ACCURACY IN STATIC CONDITION

Horizontal/Vertical Error [m]

Site 95 % Max

Rooftop (Good Condition) 2.90/5.48 4.78/9.42

Forest (Challenging Condition) 3.85/6.71 21.53/45.20

A. Static, Good Condition

Sub-meter accuracies (95% percentile) for both horizontal
and vertical coordinates were reached only with DGNSS.
EGNOS reached this level in horizontal coordinates but
vertical accuracy was slightly poorer, but still meets the Open
Fig. 4. Locations of kinematic tests. Service (OS) requirement (<4 m). The interesting observation
with SDCM is that it offers comparable performance to
Open area tests were conducted in a 3-km long road. The EGNOS, even outside its coverage area. The horizontal and
road was in the middle of fields (as shown in Fig. 5). Mean vertical accuracies are slightly worse than those of EGNOS.
elevation of obstacles was estimated to be under 10° from the
DGNSS accuracy decreases only a little when distance
horizon but still there were a few places with much higher
between reference station and rover grows. Therefore a
elevations for obstacles, for example, due to power
homogenous accuracy can be reached all over Finland. No
transmission-lines and trees. The route was driven back and
significant difference exists between the solutions offered by
forth for five times at a speed of 48 km/h with a total driving
the nearest station and the network-based DGNSS correction
time of 39 minutes.
approaches in this test environment.
City center test was conducted in the capital of Finland,
The accuracy statistics for DGNSS and SBAS systems are
Helsinki. A 50-minute driving test was carried out in more or shown in Table II and III, respectively. The horizontal scatter
less narrow streets. The average driving speed without plots in Masala rooftop and in forest environment are shown
considering any stops was 32 km/h. in Fig. 6.
During the two tests in typical Finnish road environments, TABLE II. DGNSS ACCURACY IN STATIC, GOOD CONDITION
the car was driven in a two-lane highway and in a motorway
from Raasepori to Turku and from Virolahti towards the west Nearest St. Network
of Finland. Surroundings en route varied much from open Horizontal/Vertical Error [m]
field spots to motorway tunnels and bridges. Most of the time, Site 95 % Max 95 % Max
trees and small hills were blocking lower elevation satellites.
In TD 3, the driving time was about 1 hour and 23 minutes at Metsähovi 0.35/0.65 1.25/2.41 0.34/0.61 1.36/2.48
an average speed of 75 km/h and in TD4, the driving time was Masala 0.36/0.62 1.87/4.02 0.35/0.54 1.29/3.88
about 30 minutes at an average speed of 65 km/h.
Lahti 0.50/0.75 1.34/4.84 0.40/0.68 1.64/4.86
 Fig. 6. Horizontal scatter plots in Masala rooftop (upper plots) and in forest (lower plots).

accuracy statistics for this test case are shown in Table V and
TABLE III. SBAS ACCURACY IN STATIC, GOOD CONDITION the corresponding horizontal position error plot in Fig. 7.
EGNOS SDCM
Horizontal/Vertical Error [m]
TABLE IV. DGNSS AND SBAS ACCURACY IN STATIC, CHALLENGING
Site 95 % Max 95 % Max CONDITION

Masala 0.81/1.47 2.65/4.62 1.01/1.65 4.80/9.64 Nearest St. Network

Horizontal/Vertical Error [m]
TUO2 0.93/1.65 2.81/4.71 1.05/1.59 2.96/4.13
Site 95 % Max 95 % Max
VIR2 0.97/1.72 2.43/9.66 1.31/2.01 3.35/5.24
Masala (Forest) 1.47/3.30 9.11/9.32 1.47/3.21 5.28/10.54

EGNOS SDCM
B. Static, Challenging Condition
Masala (Forest) 1.66/3.46 15.91/67.82 1.52/2.58 22.51/24.72
Positioning accuracy in challenging environment decreases
with all the tested augmentation services. But DGNSS loses
relatively the most of its accuracy, about four to six times, TABLE V. NETWORK DGNSS AND SBAS ACCURACY IN KINEMATIC,
while EGNOS and SDCM only about two times. One single GOOD AND CHALLENGING CONDITIONS
major deflection was in vertical accuracies where SDCM
Horizontal/Vertical Error [m]
reached clearly the best value. Occurrences of larger errors
became more common in challenging condition. Maximum Site 95 % Max 95 % Max
error values of DGNSS stayed the smallest. Accuracy statistics
Network DGNSS EGNOS
are shown in Table IV and horizontal scatter plots in Fig. 6.
Field Road
0.54/0.43 0.62/0.88 2.03/4.79 2.13/6.14
(TD 1)
C. Kinematic, Good Condition High-
Like the previous test cases, here also Network DGNSS /Motorway 2.91/3.61 6.18/8.99 5.90/9.42 12.48/53.03
offered better accuracy than EGNOS. During the test there (TD 3)
was a drop in number of satellites which may have at least EGNOS SDCM
partly affected the EGNOS vertical accuracy. With network High-
DGNSS, the number of satellites also dropped, but due to /Motorway 7.43/4.24 7.93/4.87 4.64/5.13 18.07/23.51
support of multi-GNSS (GPS+GLONASS), there were still (TD 4)
enough satellites visible than in the GPS-only case. The
Fig. 7. Horizontal (black dots) and vertical (red) positioning errors during Fig. 9. Horizontal (black dots) and vertical (red) positioning errors during
test drive in good condition (TD 1). Number of satellites is presented with test drive in a typical Finnish road environment in eastern Finland (TD 4).
line.

E. Availability
D. Kinematic, Challenging Condition
In challenging environments, the availability of augmented
All the tested augmentation services lose their accuracy the solutions was low due to the poor visibility of the SBAS
most in challenging kinematic condition and the offered satellites. The poorest performance was obtained in city center
accuracy from each system varies a lot during every drive. The condition. In this condition, SBAS performs poorly due to lack
best accuracy was obtained with Network DGNSS. EGNOS of corrections coming from one or two satellites. DGNSS
and SDCM performance was fairly similar. The results were solutions were then more often available as the receiver
analyzed only from drives at main roads. For the city center received corrections via the internet connection. The
condition, we did not really get enough RTK-GNSS fixed availability statistics are shown in Table VI. In Masala
solutions for reference. Thus, accuracy was not possible to be (rooftop), full availability was not achieved with DGNSS most
measured reliably with this test configuration. The accuracy probably due to a gap in the internet connection.
statistics are shown in Table V. Positioning error plots for TD
3 and TD 4 are shown in Figs. 8 and 9, respectively.
TABLE VI. NETWORK DGNSS AND SBAS AVAILABILITY

Availability [%]
Network
EGNOS SDCM
DGNSS
Masala (Rooftop) 98.08 99.99 99.54
Static
Masala (Forest) 100.00 95.40 77.11

Open Field Road (TD 1) 99.38 99.96 -

City Center (TD 2) 89.13 34.07 -

Dynamic
High-/Motorway (TD 3) 97.29 82.93 -

High-/Motorway (TD 4) - 91.41 86.09

VI. CONCLUSIONS
DGNSS solutions offer better accuracy and availability in
almost all the test environments as compared to SBAS. In
kinematic mode in good environments, the DGNSS accuracy
was almost as good as in static tests. The SBAS accuracies in
kinematic mode were not as good as compared to the results
Fig. 8. Horizontal (black dots) and vertical (red) positioning errors during
obtained in the static mode. Additionally, the Network RTK
test drive in a typical Finnish road environment from Raasepori to Tuorla (TD
3). used as a reference solution performed really well in good
signal condition.
 Availability of SBAS is limited in challenging condition [11] Kirkko-Jaakkola, M., Söderholm, S., Honkala, S., Koivula, H., Nyberg,
like in typical road environments or in urban/sub-urban streets. S. and H. Kuusniemi, 2015. Low-Cost Precise Positioning Using a
National GNSS Network. Proceedings of ION GNSS+ 2015, Tampa,
SBAS corrections come usually only from the one or two Florida, September 2015, pp. 2570-2577.
geostationary satellites which can easily be blocked by [12] Wübbena, G., “Geo++® GNSMART - GNSS State Monitoring and
obstacles. The availability of SBAS solutions can be improved Representation Technique,” can be accessed via:
further in such condition by allowing receiver to use aging http://www.geopp.de/gnsmart.
correction after SBAS signal is lost. The same notion also [13] Lenz, E., “Networked Transport of RTCM via Internet Protocol
holds for DGNSS in case of an internet connection failure. (NTRIP) – Application and Benefit in Modern Surveying Systems,” FIG
Working Week 2004, Athens, Greece, May 22-27, 2004.
The advantage of utilizing multiple GNSS systems is [14] European Global Navigation Satellite Systems Agency (GSA), “EGNOS
evident, especially when there are not enough satellites with Safety of Life (SoL) Service Definition Document,” Revision 3.0,
one single system. It is quite interesting to notice the fact that 22/9/2015.
SDCM works reasonably well in Finland outside its main [15] Russian Space Systems, “Radiosignals and digital data structure of
coverage area, though more tests are required to understand GLONASS Wide Area Augmentation System, System of Differential
Correction and Monitoring,” Interface Control Document, Edition 1,
what level of accuracy we can expect from SDCM in different 2012.
parts of Finland. The same goes also for EGNOS.

ACKNOWLEDGMENT
This research has been conducted within the FEGNOS
(Finland’s EGNOS Monitoring and Performance Evaluation)
project, funded by the Finnish Transport Agency and the
Finnish Geospatial Research Institute. We would like to thank
Mr. Antti Laaksonen for driving preliminary tests, Mr. Ari
Huvinen and Mr. Marko Halmelahti for their help to make test
measurements possible in Lahti, Dr. Kaj Wiik and the other
staff in Tuorla observatory for their help with practical
matters.

REFERENCES
[1] European Global Navigation Satellite Systems Agency (GSA), “EGNOS
Service Definition Document – Open Service (OS),” Version 2.2, can be
accessed via: http://egnos-portal.gsa.europa.eu/library/technical-
documents.
[2] The European Satellite Services Provider (ESSP), available online via:
http://egnos-portal.gsa.europa.eu/discover-egnos/programme-
information/status.
[3] “Averin, Sergey V., Dvorkin, Vjacheslav V., Karutin, Sergey N.,
"Russian System For Differential Correction And Monitoring: A
Concept, Present Status, And Prospects For Future,” Proceedings of the
20th International Technical Meeting of the Satellite Division of The
Institute of Navigation (ION GNSS 2007), Fort Worth, TX, September
2007, pp. 3037-3044.
[4] Gibbons, G. “Russia Building Out GLONASS Monitoring Network,
Augmentation System,” Inside GNSS, Issue: September/October 2009.
[5] Solberg, A., Hanssen, R. “Monitoring EGNOS in Norway,” NKG
General Assembly 2014, Göteborg, Sweden.
[6] Jensen A. J., Hanssen R.I.,Mikkelsen A., de Mateo J., “EGNOS
Performance in Northern Latitudes,” Proceedings of the 13th IAIN
World Congress, Stockholm, 27-30 October 2009. Published by the
Nordic Institute of Navigation.
[7] PEGASUS Software User Manual, EUROCONTROL, version 01-Q,
29/09/2003.
[8] Finnish Geopsatial Research Institute, “Finland’s EGNOS Monitoring
and Performance Evaluation (FEGNOS)”, can be accessed via:
www.fegnos.net.
[9] Finnish Geopsatial Research Institute, “The DGNSS Service from
Finnish Geospatial Research Institute, National Land Survey of
Finland,” can be accesed via: http://euref-fin.fgi.fi/fgi/en/positioning-
service/dgnss-service.
[10] Koivula, H. et al., “Finnish Permanent GNSS Network,” Proceedings of
the 2nd International Conference and Exhibition on Ubiquitous
Positioning, Indoor Navigation and Location-Based Service (UPINLBS
2012), 3–4 October 2012, Helsinki, Finland. IEEE Catalog Number:
CFP1252K-ART. ISBN: 978-1-4673-1909-6.