Beruflich Dokumente
Kultur Dokumente
O-8725
Frequency Management Working Group
Title: GSM-R MS Interference measurements at Ispra,
December 2011
Date: April 02, 2012
Version 2.0
Status: Final
Pages: 26
Name Organisation
The authors want to express their special appreciation to the Ispra Joint Research Centre, and
especially Mr Gianmarco Baldini and his colleagues, for making the measurements possible.
1 Management summary
In November 2011, the UIC Frequency Work Group decided to organise an interference test
campaign. Using market available, real GSM-R radios, a set of objective measurements would be
performed to demonstrate the real blocking behaviour of GSM-R radio modules caused by 900MHz
broadband specifically UMTS and LTE, downlink signals. These tests were done in week 50, 2011, at
the European Commission’s Joint Research Centre (JRC) in Ispra, Lago Maggiore, Italy. This is a fully
accredited test lab, which is also used for e.g. ETSI and CEPT radio test campaigns.
A total of 4 ETSI compliant GSM-R radios have been used in the tests. Our test results show the
following1:
For the current GSM-R radios, a potential method to improve the capability of GSM-R radios to resist
interferers is to use protection filters limiting the interfering signals entering the equipment. Our tests
show that such protection filters must have sufficient rejection properties for interfering signals close to
the GSM-R downlink band. It is also important to maintain the rejection across the public GSM band.
Note that unwanted emissions from UMTS or LTE base stations to the GSM-R downlink, which are
additional factors causing interferences to GSM-R, were outside the scope of the current tests; these
may be subject for further testing.
Based on the above test results and seeing the developments in the 800 & 900 MHz frequency
environment, UIC proposes to study the following items related to GSM-R that so far have not
sufficiently been covered within CEPT ECC reports:
1. Consider the real interference behavior of receivers of standards compliant GSM-R
terminals when exposed to broadband signals. This includes inter-alia:
Receiver blocking capabilities;
Impact from aggregated (composite) interferer power. UIC notes that this problem may
also affect other radio services and applications such as wireless short range
communications or public mobile communications;
Dependency from interferer frequency offset;
2. The impact of LTE(800) operating in the band below 862 MHz on GSM-R. UIC notes that
some studies are already ongoing in the ECC WGSE and GSM-R is not covered yet.
1
The purpose of this report is to make the UIC test results available. The reader of the report is urged to analyze
the impact of these tests on assumptions used in various previous studies.
3.1 Participants
The tests have been prepared and executed by a small team of experts, identified at the UIC
Frequency workgroup meeting of November 8th.
For these tests, GSM-R radios from different manufacturers have been used. In order to avoid any
commercial conflict of interests for the radio module manufacturers, all testing has been performed
under UIC NDA, and all test results have been anonymized. Each of the participating manufacturers
has received the detailed test data of their own equipment only. The final test report, with the
anonymized results, will be made available to all members of the FWG and all radio manufacturers.
The purpose of the tests was not to evaluate the relative performance of the different manufacturer’s
radios, but to measure and understand their blocking behaviour and mechanisms. Selection criteria for
the radios to be used for the testing were:
Commercially available:
Compliant to ETSI TS 100 910: "Digital cellular telecommunications system (Phase 2+); Radio
Transmission and Reception (3GPP TS 05.05 Release 1999)", or
Compliant to:
ETSI TS 102 933-1 V1.1.1 (2010-12) Railway Telecommunications GSM-R improved receiver
parameters; Part 1: Requirements for radio reception, and Part 2: Radio conformance testing.
It should be noted that the testing has been focussed on blocking due to interferers in the 900MHz
public operator band. No testing has been done with respect to interferences due to unwanted
emissions, nor due to the Digital Dividend 800MHz band; these may be subject for further testing.
In chapter 4 of this document, receiver blocking measurements and results are provided. Chapter 5
then demonstrates the effects of different blocking protection filters. And in Chapter 6 information
regarding the test set up is provided, and the conclusions from all tests derived.
Note that unwanted emissions from UMTS or LTE base stations to the GSM-R downlink, which are
additional factors that cause interferences to GSM-R, were outside the scope of the current tests.
All measurements were performed using standard test equipment, mainly available from the Ispra JRC
lab. Instead of a GSM-R base station, a Rohde & Schwarz base station simulator was used. This
enabled reading the RxLev and RxQual parameters as reported by the mobiles, independent from the
mobiles themselves.
As the measurements were focused at the effects of unwanted phenomena and IM products causing
blocking inside the radios under test, it was of utmost importance to ensure that the generated test
signals were sufficiently clean from unwanted emissions. This was achieved by proper usage of
isolators and the usage of a GSM-R band rejection filter (implemented as a band pass filter + hybrid
combination) to reject the out-of band emissions from the interference generators. Filter type 1 as
defined in section 5.4 was used for this.
All test equipment used was properly calibrated. All measured signal levels were referred to the input
connector of the GSM-R MS. Further details of the test set-up are provided in section 6.1.
For all tests, the GSM-R MS was in dedicated mode on a GSM-R channel, using a normal full rate
voice channel. The interfering signal levels were increased until the tested GSM-R MS changed its
reported GSM quality class from 4 to 5. This quality class change from 4 to 5 corresponds to 3,2 % bit
error rate, and is the point where a standing GSM-R call is dropped.
The purpose of this test is to verify the capability to resist blocking when there is a single UMTS
interferer, using different GSM-R MS receive levels.
The test-setup according to Figure 1 was used, with one interfering UMTS signal at 927.6 MHz. The
wanted GSM-R downlink signal level was increased in 10 dB steps, starting at -101 dBm.
Three different types of mobiles called Mobile 1, 2 and 3 were tested. The results for each mobile are
provided in the following sections.
4.2.1 Mobile 1
Comments of results
Note the near third order behaviour of the blocking performance of this mobile. For example, between
level – 101 dBm and – 71 dBm a difference of 30 dB of wanted signal for channel 964, the interfering
signal only increases from – 25 dBm to – 15 dBm = 10 dB.
This property is typical of the blocking type (intermodulation of the third order) experienced with
broadband signals of the UMTS / LTE etc type.
From this and the following single signal blocking results, it is clear that increasing the wanted signal to
counter the interference causing blocking is very expensive (build extra base stations to increase
GSM-R signal levels).
Also note the relative small difference of blocking performance (resistance to blocking) of about 5 dB
between the lowest (955) and highest (973) GSM-R channels. Wanted signal levels higher than –31
dBm were tested. However, the tested terminal then failed reporting a reliable RX-level.
Mobile 1 exhibits a somewhat better blocking performance compared to Mobiles 2 and 3 (see pictures
below).
This mobile is of a new generation GSM-R mobiles with improved linearity and selectivity according to
the ETSI specification TS 102 933-1 V1.1.1 (2010-12).
Note, as with Mobile 1, the near third order behaviour of the blocking performance of this mobile. Also
note the relative small difference of blocking resistance of about 5 dB between the lowest (955) and
highest (973) GSM-R channels, and almost non-existent difference between mid and high GSM-R
channels 964 and 973.
Wanted signal levels higher than – 41 dBm were tested. However, the tested terminal then failed to
report reliable RX-levels.
Note, as with Mobiles 1 and 2, the near third order behaviour of the blocking performance of this
mobile for wanted GSM-R signal levels up to -71 dBm. Above this level the graph shows a steep (it
was in fact a step) improvement in interfering signal handling. However, this improvement caused a
drop in wanted GSM-R signal sensitivity.
When lowering the wanted GSM-R level a few dB’s below -71 dBm without the interfering UMTS
signal, the terminal reported RXqual 4-5 and eventually dropped the call. The vendor explained this
behaviour as a software mistake. Another unit (Mobile 3, Unit 2) of the same type but with different
software was then used for further testing.
Note that only the highest GSM-R channel (973) is shown in this blocking performance test. This
mobile unit also show the typical near third order blocking behaviour like mobiles 1 and 2. Wanted
GSM-R levels higher than – 31 dBm were tested with similar RX-level reporting problems as with
mobiles 1 and 2.
The purpose of this test is to measure the blocking effects when using different downlink fading
profiles and different terminal speeds in the wanted signal. The interfering signal was a static, single
UMTS signal at 927,6 MHz. The wanted GSM-R signal (without fading) was set at level – 71 dBm at
channel 973. The following test set-up was used:
Note the limited but clear influence of fading on the GSM-R downlink blocking resistance with a single
UMTS interferer. Usually the measured effect of fading was in the range 2…3 dB compared to static
conditions. This effect of dynamic signal conditions should be taken into account when maximum
interfering signal levels are defined. The limited influence is a result of the efficient channel coding /
interleaving of GSM.
From the results on blocking properties of the three different GSM-R terminals it is clear that there will
be severe interference problems from high level broadband signals in the 900 MHz Public Mobile
Operator band.
As a practical example, the Swedish regulator, PTS, has issued frequency licenses with signal power
limitations for the Public Operators. The maximum signal level for the directly adjacent UMTS carrier,
i.e. in the 925 – 930 MHz spectrum, to be handled by the GSM-R mobile installation is – 5 dBm / 5
MHz (defined as the maximum received power into a 0dBi antenna 4 m above the rail).
The best blocking test result of the three different GSM-R terminals for the spectrum in question is
seen in for Mobile 1 in figure 2. When exposed to a -5dBm UMTS signal, in order to have a good
GSM-R signal quality, this mobile must see a wanted GSM-R signal of around (-36 +4) 2 = -32dBm
without fading, if operated on channels near the GSM-R band edge (973). Such a high signal level is
of course very difficult to create in any practical GSM-R network 3.
The purpose of this test is to understand the interference effects of a broadband interfering signal as
this was shifted across the E-GSM downlink band 925 – 960 MHz.
For one mobile, this test was done using 200 kHz steps from 927.6 MHz to 932.6 MHz, as well as with
5 MHz steps from 927.6 to 957.6 MHz.
Another mobile was tested only with 5 MHz steps from 927.6 to 957.6 MHz
2
At -36dBm the GSM-R call is dropped. In order to not drop the call, and have a good signal quality (i.e. RxQual
= 0), the GSM-R signal must be 4dB stronger, thus -32dBm.
3
The worst interference case is that where the wanted GSM-R signal is at its weakest, so at the hand-over
points. There, the network will have to (at least) comply with the Eirene coverage lower limit (e.g. -95dBm). It is
not realistic to create a signal of -30dBm or so at those points. But also at any realistic distance from a GSM-R
BTS such high levels will be difficult to achieve. But UMTS or LTE base stations / interferers can be quite close to
the victim GSM-R receiver, so quite easily can create levels of -30dBm or higher at the victim.
Figure 8. Mobile 1 selectivity towards a UMTS interferer over 927,6 to 932,6 MHz in 200 kHz steps
Note the relative small (about 6 dB) selectivity towards the 5 MHz UMTS interferer when it is shifted 5
MHz in 200 kHz steps away from the GSM-R channel. Introducing a “Guard-band” towards GSM-R by
moving the first UMTS channel to 928,8 MHz as suggested in some compatibility studies will not be of
much help for this case of blocking.
Figure 9. Mobile 1 selectivity towards a UMTS interferer over 927,6 to 957,6 MHz in 5 MHz steps
Comment; Mobile 1 exhibits better selectivity towards a 5 MHz UMTS interferer compared to Mobiles 2
and 3 (see the next results).
Comment; Note that selectivity of this mobile towards a 5 MHz LTE interfering signal is similar to the
UMTS interfering signal case.
Figure 11. Mobile 1 selectivity towards a 10 MHz LTE interferer over 930 to 950 MHz in 10 MHz steps
Note that the nearest GSM-R channel (973) seems to have better selectivity towards the broadband
10 MHz LTE signal. This channel would be expected to be less resistant to intermodulation-like
interference generation than the other channels further down the GSM-R band. Instead there seems to
be other interference mechanisms involved in this case.
Figure 12. Mobile 2 selectivity towards a UMTS interferer over 927,6 to 957,6 MHz in 5 MHz steps
Comments to fig 12; This terminal is a new GSM-R version according to the ETSI TS 102 933-1
V1.1.1 (2010-12) specification with better linearity and selectivity than the GSM 05.05 requirements.
4.4.3 Mobile 3
Figure 13. Mobile 3 selectivity towards a UMTS interferer over 927,6 to 957,6 MHz in 5 MHz steps
The selectivity of the tested mobiles to interference from broadband signals anywhere in the public
GSM 900 MHz band is not sufficient. This is explained in the example case below, based on the
current situation in Sweden:
The Swedish PTS license limitations for the Public Mobile Operators is 0 dBm /5 MHz in the
930 – 960 MHz (defined as the maximum received power into a 0dBi antenna 4 m above the
rail).
Assume a public mobile operator using a 5 MHz UMTS signal at 932.6 MHz using the power
of 0 dBm.
Further assume simple third order blocking generation as interference mechanism.
Even the best mobile in this test, Mobile 1, must see a wanted GSM-R signal of -38 dBm or
more for good signal quality.
The second best mobile requires – 20 dBm for good signal quality.
These levels are high and very difficult to achieve. Also they will only handle interference from
one (1) single broadband signal at the example frequency.
It is obviously necessary to protect GSM-R terminals from harmful interference from the public GSM
band. The selectivity must be significantly improved in order to correctly handle the wanted GSM-R
signals.
Based on the current measurements results, the required selectivity, against broadband interferers,
can only be (realistically) realised through protection filtering of the wanted GSM-R signal spectrum.
The purpose of this test is to understand the effects of multiple (in our tests 2) interferers, with the 2 nd
interferer stepped in frequency in order to understand the effects of frequency separation. The test-
setup according to Figure 1 was used, with two interfering signal sources.
In this test, two interfering UMTS signals were used, one fixed at 927,6 MHz, the first recommended
UMTS channel for the UMTS 900 band, and the other stepped across the public GSM downlink band
in 5 MHz steps from 927.6 to 957.6 MHz.
Levels of the interfering signal sources were set equal and 3 dB reduced in order to generate the
same composite power as the single interferer setting for generating RXqual 4-5 as shown in Fig 2 for
927.6MHz.
Comment; When both interfering UMTS signals are at 927,6 MHz, Mobile 1 sees these as a composite
power and performance is the same as with one UMTS carrier at the same frequency. When one
UMTS carrier is stepped towards higher frequencies the composite power limits the selectivity of the
mobile.
This composite power property of the mobile is important to understand.
Comparing the Mobile 1 blocking property for a single interferer at 952.6 MHz around – 1 dBm (see
fig. 9) with the composite result of – 18 dBm shown in the above graph, clearly indicates the impact of
composite power over the whole interfering spectrum.
If a national frequency regulator issues power limitations for the lower part of the public GSM
spectrum, he cannot ignore the composite effect when setting the power limits for higher frequencies
in the licensed spectrum. This means that coordination will be required between the GSM-R operator
and all public operators within the 900MHz frequency band, as well as between all public mobile
operators in the 900MHz band..
In this test, using Mobile 1, one interfering UMTS signals was used, fixed at 928,8 MHz. The other
interferer was a narrow-band (CW) signal, placed in the spectrum between GSM-R and the broadband
signal. The narrowband signal was stepped between 925,2 and 926 MHz in 200 kHz steps.
The graph below shows the level of each interferer (UMTS and CW). The composite power of both
interfering signals was consequently 3 dB higher (-15,2 dBm).
Comment; This test signal constellation generates through the GSM-R receiver non-linearity a
broadband (UMTS band width) interference product that falls into the GSM-R down-link band.
Compare the single UMTS interferer blocking level at 928,8 MHz (see figure 8 above around – 15 dBm
with the composite level [( - 18 dBm)+ (- 18dBm) = - 15 dBm] in this test. Placing a narrow band signal
(i.e. GSM) between the broadband signal such as UMTS and GSM-R clearly is not a good proposition.
4.5.3 Conclusion of blocking from several broadband signals in the 900 MHz Public band
Selectivity of the tested mobiles towards interference from several broadband signals anywhere in the
public GSM 900 MHz band is not sufficient.
The selectivity is very limited when combinations of several wide/wide or wide/ narrow signals are
allowed to enter the GSM-R terminal. Even the best tested mobile, Mobile 1, can only handle a limited
composite interference power.
If the tested case with one UMTS interferer at 927,6 MHz is combined with a similar interfering signal
anywhere in the public GSM band, only around –16 dBm composite power can be handled with
RXqual 4-5 when the wanted GSM-R signal is operating at channel 973 with –71 dBm input signal
(see figure 14).
As a theoretical example assuming near third order behaviour for the above (+ 8.0 dBm) interferer
level, the wanted GSM-R signal level must be 4 – 71 + [ 8.0 – (-16 )] x3 = + 1 dBm. To have an RXqual
of 0, (i.e. good signal quality) the wanted GSM-R signal must to be at least 5 + 1 + 4dB = +5 dBm !. The
previous tests show that no GSM-R terminal is operational at this input signal level.
It is clearly important to protect GSM-R terminals from harmful interference from composite power
effects from the public E-GSM band. The selectivity must be greatly improved to handle only the
4
Explanation: -71 = wanted GSM-R signal level used in the test of Fig 14. To this the difference between the
interference level used in Fig 14 and the Swedish +8,0dBm needs to be added, multiplied by 3. The factor x3 is
due to the 3rd order behavior as shown in Fig 2 etc.
5
Explanation of the 4dB: see footnote 2 at page 12
This test was performed to show the capability of the GSM-R radio to recover after dropping a call,
from a blocked situation when the level of the interference is lowered (i.e. train is moving away from
the interferer). For this test, the interferer signal was reduced until the radio was able to re-establish
the call.
Tests were performed on Mobile 1 operating at channel 964 (mid GSM-R band) and at a wanted level
of – 71 dBm:
Blocking was defined when RXqual 4-5 (loss of service) was experienced. This happened at
interferer level of -15,1 dBm;
After the call had dropped, it was tried to establish a new call using lower interferer levels;
The new call was possible with a RXqual = 2 at interferer level of -16,3dBm.
Comment: It only takes 1.2 dB increase of the wanted signal level to regain service after being
blocked.
In order to demonstrate the effect of protection filters, as a function of their filter characteristics,
blocking tests were done with five different protection filters 6. The transmission plots of these different
filters are provided below.
The test setup as Figure 1 was used, with the protection filter inserted ahead of a GSM-R Mobile 3
unit.
The blocking performance was recorded for two GSM-R wanted signal levels, - 101 and – 71 dBm.
The interferer was a single UMTS (5MHz) signal at 927.6 MHz, set to a level were the mobile reported
RXqual 4-5.
6
The Passband characteristics of these filters are not considered relevant for the purpose of the currents tests,
and therefore not elaborated in the following graphs.
Note that filters #2, #3 and #5 do not remove all of the 927.6 MHz UMTS signal content (see plots and
tables below) with the result that blocking of the intermodulation type is generated in the GSM-R
terminal.
To indicate the usefulness of the tested filters, the following examples are based on the current
situation in Sweden..The Swedish Regulator, PTS, has defined power limitations for the public
operator for the 925 – 930 MHz spectrum (with the typical UMTS centre frequency = 927,6 MHz as
used in these tests). For the spectrum in question, PTS has set the limit to – 5 dBm / 5 MHz.
Filters #2, #3 and #5 show no or limited improvement of the blocking performance (see blocking
results of Mobile 3, unit 2) in this test. With these filters used in the PTS case above, the calculated
necessary GSM-R level for RXqual 4-5 must be at least 7 –101 + (3 x 32) = -5 dBm. It is clear that
these filters are of no use in the Swedish case.
The plots of Filter #1 and #4 indicate that there should be little or no problems with blocking of
intermodulation type. The blocking at – 13 and – 15 dBm is probably not caused by intermodulation.
Instead there is a high probability that unwanted emission from the UMTS interference generator
together with limited attenuation of the rejection filter (see Figure 1) at 924,8 MHz (channel 973) is
seen in the plots.
With higher wanted GSM-R signal level as in the plots for – 71 dBm below, the unwanted emission
levels are not high enough to interfere with the results of blocking of intermodulation type.
7
Explanation: -101dBm is the wanted GSM-R signal level. -32dBm is the interferer level when using filter #3
when Rxqual is reduced from 4 to 5, so the call is dropped. The multiplier 3 is due to the 3rd order behaviour.
Note that filter # 1 and #4 appears to allow an interferer at 0 dBm. However, as the output power of the
interference generator approaches + 20 dBm (maximum output for UMTS), the UMTS signal is likely
to be clipped, resulting in reduced Peak to Average Ratio, and less blocking.
It is important to have enough rejection at all spectrum parts with interfering power. Using a mobile
with blocking properties as the tested Mobile 3, filters #1 and #4 could be used in the Swedish PTS
case. Filters # 2, #3 and # 5 will not provide necessary blocking improvement in the same case.
Below are the plots of the protection filters. Filter #1, #2 and #4 are of dual bandpass type for the
GSM-R up-and downlink bands. Filter #3 is a single bandpass filter with a steep slope towards the E-
GSM band. Filter #5 is a single bandpass filter with a moderate slope towards the public GSM band.
A few important attenuation points have been shown for each filter in the tables below. The 925,5 MHz
point is important as this is where the power of a UMTS signal operated at 927,6 MHz with a 4,2 MHz
power bandwidth ( 99,5 % ) starts to drop off. High attenuation from this drop-off point and further up in
frequency is important.
Filter #2
Filter #4
The testing reported in this document is part of a measurement campaign initiated by the UIC
Frequency Work Group. The main subject for this phase of the campaign has been to verify some
blocking properties of the existing GSM-R receiver, when subjected to broadband (UMTS, LTE)
interferers. The blocking measurement results have been obtained with a set of tests on a small
number of ETSI compliant GSM-R terminal units, currently available on the market.
It should be noted that unwanted emissions from UMTS or LTE base stations to the GSM-R downlink,
which are additional factors causing interferences to GSM-R, were outside the scope of the current
tests; these may be subject for further testing.
The mobile units used for the tests, complied with either:
ETSI EN 300 910 V8.5.1 (2000-11) GSM 05.05 version 8.5.1 Release 1999, or
ETSI TS 102 933-1 V1.1.1 (2010-12)).
Signal levels:
All signal levels were referred back to the input connector of the mobile under test. Except for
the interferer signal in the filter tests, in which case the interferer level was referred to the input
connector of the filter, and the wanted signal to the input connector of the mobile;
Signal powers for GSM-R were measured over a 200kHz bandwidth;
Signal powers for UMTS or LTE were measured over a 5MHz or 10MHz bandwidth;
In many tests – 71 dBm was used as the wanted GSM-R level. One reason for this is to have
the interference generation of the GSM-R terminal only from intermodulation as this is the
predominant blocking mechanism. Another reason is that the level is in the middle of -101 to –
41 dBm span used in the amplitude domain tests (see para 4.2 Single UMTS interferer at
different GSM-R levels).
Level calibration of the entire test set-up as per figures1 and 6 were performed using a network
analyser. All measurements reported in this document have been appropriately corrected based on
these level calibrations.
An important result of the tests is the blocking behaviour in the amplitude domain, using one interfering
signal.
All but one test were performed using static levels of both interferer and the wanted GSM-R signal.
The additional test that introduced fading on the GSM-R downlink signal shows degradation of the
GSM-R terminal interference rejection (blocking) capability. Even if that degradation is relatively small
(up to 3 - 4 dB), it must be included when designing interference protection measures.
For a single interfering signal, the blocking behaviour in the frequency domain shows better
interference rejection performance of the GSM-R terminal as the frequency separation between the
GSM-R wanted signal and the interferer is increased. This was expected.
But when several (in our measurements 2) interferers are used, the rejection is shown to be limited.
This result is an indication that it is the composite, or aggregated, interference power that will severely
limit the GSM-R receiver capability even if the interferers are well separated in frequency from the
GSM-R band.
Given the magnitude of the blocking effects described in the previous sections, and current
commercial & technical constraints, we consider it highly unlikely that any existing or future developed
GSM-R terminal can provide sufficient resistance to blocking from signals in the public mobile
operators 900 MHz band.
A potential method to improve the capability of GSM-R radios to resist interferers is to use protection
filters, limiting the interfering signals entering the equipment. Our tests show that such protection filters
must have sufficient rejection properties for interfering signals close to the GSM-R downlink band. It is
also important to maintain the rejection across the public GSM band. It is important to develop a
general filter specification. However, this can only be done after more complete testing (see below).
It should be noted that the tests performed do not provide full understanding of all parts of the blocking
interference mechanisms. Information from other spectrum users, specifically Air Traffic Control at
frequencies above the public GSM band, indicate that additional, more complex testing is necessary,
using a multitude (perhaps 10 – 12) of interfering transmitters. Therefore, a more elaborate
characterisation of the GSM-R terminal blocking behaviour to these types of interferences is still
necessary.
Only when this has been completed is it possible to set power limitations for the public operators and
also design necessary protection measures for the GSM-R use.
The current testing has been done with interferers in the public mobile 900MHz frequency band.
Interference problems from “Digital Dividend” 800MHz band or other public mobile (or fixed) users of
spectrum below the GSM-R band are not well understood. Of particular concern in this frequency band
is the interchange of uplink and downlink frequencies, compared to that in the 900MHz band. With the
potential volume of 800MHz band broadband user terminals (mobile and fixed) in or nearby railway
stations, having limited unwanted emission requirements and unpredictable locations, it is highly
probable that these will also create interference problems to the GSM-R services. This area clearly
requires further study by the UIC railway community.