REFERENCE O-8751-1.0 FREQUENCY MANAGEMENT WORKING GROUP Input to FM54 on OOB emissions due to UMTS or LTE signals NAME DATE VISA Author FM Drafting Group 10/2014 D. Martens Reviewed Endorsed FM WORKING GROUP Chairman GSM-R ERIG Group Chairman 10/2014 D.Schattschneider 10/2014 R.Sarfati O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 1 of 9
EVOLUTION SHEET Version Date Author Modification 0.1 October 2014 FM Working Group Draft edition 0.2-0.5 October 2014 FM Working Group Draft editions, incorporating various FM WG inputs 1.0 October 2014 FM Working Group Final edition O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 2 of 9
CONTENTS 1 INTRODUCTION 4 2 USE OF STANDARDS 4 3 CALCULATION BASIS 4 4 C/I REQUIREMENTS 5 5 OOB CALCULATION MODEL 6 6 HAND-OVER MARGINS 6 7 DESENSITIZATION 7 8 TELEFONICA CONTRIBUTION 7 9 ANTENNA FACTOR 7 10 REFERENCES 9 O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 3 of 9
1 Introduction This document has been prepared by the UIC Frequency Management Working group to provide input to the CEPT/ECC FM54 work on providing guidelines for achieving co-existence between GSM-R and MFCN networks that are using UMTS or LTE in the 900MHz band. In the FM54#2 meeting held in September 2014 in Copenhagen, a number of discussion items have been raised where differences appeared to exist between the UIC view and that of the MFCNs. This document aims at clarifying these items by providing additional background information to the UIC views. It is to be used in the FM54 web conference planned for November 5th, 2014. This document addresses the items raised due to the FM54#2 meeting in separate chapters 2 to 9. As these items are mere input elements for the FM54 report, this document cannot draw any conclusions. 2 Use of standards As a principle, the UIC position is that for the FM54 report all calculations related to radio performance shall be based on parameter values as defined in the applicable ETSI or 3GPP standards. These standards have been agreed and accepted by all parties. Also, they provide clear and well understood reference points. Other values such as measured or practical values potentially create less clarity and different understandings, which is to be avoided for reports. It is the UIC view that for the FM54 report only parameter values as defined in relevant standards should be used. 3 Calculation basis During the FM54 Copenhagen meeting, the subject of ability to measure a 95% coverage probability was raised. Also the question was raised whether calculations for OOB levels should be based on 95% or 50% values. The UIC view on this is as follows: The EIRENE SRS requirement defines that GSM-R networks shall have a coverage probability of 95% over each 100m section of railway track In order to verify this, GSM-R network operators measure the coverage level using a high number of samples (one sample is calculated (as local average or RMS value) for e.g. each 7,5cm, 10cm, or 36cm) Of all those samples, it is verified if over a 100m section the 95% value is equal to or higher than the required coverage level, e.g. higher than -98dBm (for voice) 1 This method thus is not based on the Lee criterion, and in fact is a more stringent test as the averaging over a distance of 40 lambda is avoided. Samples that due to fast fading fall below the required coverage level are included in the 95% calculations, and not averaged out In view of the above, in general radio signal parameters should be defined as 95% values. If a 50% value is needed, this can be derived from the 95% value via the Cumulative distribution function. This applies to both the carrier (C) level, and the Interferer (I) level. As C/I is a pure ratio, both parameters should be based either on 95% or 50% values, in which case it does not matter which percentile is used. It is to be noted that the standard deviation of radio propagation models differ for the various GSM-R networks (e.g. mountainous versus flat terrains). 1 Some GSM-R networks use a somewhat different measurement methodology; some apply the Lee criterion with averaging over 40 lambda, some use floating 100 m intervals, etc.. O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 4 of 9
The UIC proposes to base the FM54 report on the usage of 95% values for coverage and interferer levels. These values then may be converted to average (50%) values, for example to be used in radio planning tools, by applying a radio margin based on the standard deviation depending on the radio propagation model applicable to the specific network. Likewise, the minimum coverage level then may be adjusted for the actual design level for the specific GSM-R network Some examples of the radio margin that has been / is being used in the design of several GSM-R networks are: o France: 10dB o Netherlands: 13dB o UK: 15dB o Germany: 11 db o Belgium: 12 db o Please note, that these radio margins are directly linked to the standard deviation of the propagation model and to the accuracy of the terrain date base. 4 C/I requirements During the FM54 Copenhagen meeting, the question was raised what C/I value is required for GSM-R services. The UIC has the following view on this: As the UIC starting point is always the values defined in standards, a C/N of 9dB is to be used for voice as implied in EIRENE and 3GPP 45.005: -121dBm is the thermal noise at the receiver input in a 200kHz bandwidth, plus 8dB noise figure of the receiver gives -113dBm as the receiver noise floor. The receiver reference sensitivity is -104dBm, thus the C/N is 9dB. This ratio should not be reduced when introducing interferences, thus C/(N+I) = 9dB, where the I is the sum of co-channel, adjacent channel and OOB interferences. Measurements on actual equipment could show that some pieces of equipment perform better than this, but such values are not guaranteed by the manufacturers, and thus cannot be applied to further calculations The measurements done at Bundesnetzagentur indicate a C/I value of 6 db, corresponding to an RXQual value of 4, which was used as the measurement stop criterion. RXQual 4 allows a max. bit error of 3,2 % in an ongoing call, which is more robust than the processes going on during the initiation of a call. The network design target however is for good quality, uninterrupted communication via GSM-R, so with Rxqual = 0 or 1. The difference between an RXQual value of 0 and 4 then is available to improve the GSM-R system robustness (e.g. for performance variations between trains, seasonal propagation variations, external interferences). In other words, the GSM-R networks are not designed to normally operate at an Rxqual = 4 or higher. It is to be noted that for other services (CSD, GPRS) higher C/I values are required (see also the ongoing activities in ETSI for a packet (GPRS) based ETCS). Also, on national level higher C/I requirements may be used to cover specific national needs For the discussion on OOB emissions, we consider the OOB effects due to UMTS or LTE carriers to be noise like, and therefore do not need to be concerned with the question of correlation between the wanted and interfering signals (i.e. is similar to thermal noise) Since C/(N+I) is given as a ratio, it is independent of the used values being defined as of 50% or 95% values, as long as all are of the same type 2 For all C/I related discussions, it is to be noted that the EIRENE minimum coverage requirement of -98dBm is directly linked to the ETSI defined receiver reference sensitivity by the EIRENE defined 6dB loss between 0dBi train antenna and the receiver module input connector. Thus the 9dB C/I requirement is equally valid at the receiver input as well as in the air. 2 Note that C/I measurements are always showing the instantaneous values of the C and the I components, which subsequently may be mathematically converted to e.g. an average value. O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 5 of 9
5 OOB calculation model The attached Excel spreadsheet was originally proposed by UIC at FM54#2, and somewhat modified during that meeting. The UIC agrees with the calculation principles used therein, but with parameter values to be used as discussed in the other sections of this document. The following changes have been made to the attached version of this calculation spread sheet: Removed the rows that calculate the maximum UMTS carrier level. As the purpose of the model is to manage and therefore calculate the allowable OOB levels, the UMTS carrier level is not relevant Removed the tab called average, When calculating based on 50% values, an additional hand over margin must be included, Thus the tab called average HO is sufficient and retained. Added comments to sheet Minimum rows 17 and 20 Added to sheet Average HO rows 6 and 7 to handle the conversion from 95% to 50% values Added to sheet Average HO row 8 to handle the hand-over margin Added a comment column to both sheets 6 Hand-Over margins When discussing the OOB calculation model, it was suggested that an additional hand-over margin should be applied to the minimum coverage level. Further investigation has however resulted in the understanding that, when coverage is based on 95% values, no additional handover margin needs to be considered. However when using 50% values, a handover margin of 5dB is necessary. The Handover is a key function in the GSM-R radio network and the handover margin is a fundamental part of the handover function. The handover margin will be used for the calculation of the handover decision by the BSC and will avoid subsequent handovers (also known as Ping-Pong Handover effects). A handover will only be carried out when the reception level of the target cell is received with a level that is greater than the total sum of the reception level of originating cell added with the handover margin. High handover margins can result in dropped calls, while very low values of handover margin can produce Ping-Pong effects, as a MS switches too often between cells. The value of the handover margin is set individual per cell neighborhood (the relation between source cell and target cell ). The following figure shows the effect of the handover margin in the handover area between two GSM- R radio cells: O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 6 of 9
7 Desensitization In discussions at the FM54 Copenhagen meeting, it has been stated that some ETSI / 3GPP documents define the amount of desensitization allowable for mobile stations. However, for the documents identified in this meeting, UIC can only identify the following: 3GPP TR 43.030, whose purpose is to provide guidance on cell planning, does not define an allowable level of desensitization. Only in section D.11.1.1 the subject of desensitization is mentioned in relation to a GSM base station, but it does not define any level for this. Actually this section only defines a 3dB noise margin, resulting in a desensitization lower than 3dB (but unspecified) 3GPP TR 45.050 where background information on GSM RF requirements is provided. Here desensitization (of 3dB) is used in the context of defining a limit for receiver blocking (paragraph U.3.3.1) The LS sent by ECC PT1, referenced PT1(12)084/SE7(12)061, where a 3 db margin is defined for the UE. This document states a 3dB desensitization for a MFCN mobile station. However, this is based on the CEPTRep030, which does not provide any justification for such a 3dB number. On the contrary, it actually uses a 1dB desensitization for the victim receivers considered, i.e. the DTT receivers. In view of this, UIC do not see any justification to use a 3dB desensitization factor for UE in the ER-GSM band ETSI discussions, only 0,8dB desensitization of the base station has been allowed Based on the above, UIC see no reason to use a desensitization factor for the GSM-R MS higher than 1dB in OOB related calculations. 8 Telefonica contribution UIC have reviewed the Telefonica contribution UMTS 900 vs GSM-R Study draft 0.7.pdf, which is using the Seamcat modeling tool, and have the following remarks: It is not clear for which distance between the MFCN base station and the rail track the calculations have been performed. For existing networks with interferences by MFCN base stations, these can vary from as little as 20m to several hundred meters or kilometers. Obviously, the closer distances are the more interesting cases. According to Figure 1 of the study, the minimum distance between UMTS 900 DL Tx and GSM-R DL Rx seems to be ca. 500m. As the used Seamcat model is based on using 50% values, a handover margin of 5dB is necessary. This however has not been used in the calculations. Calculations must be related to the interference impact over 100m sections, and not as averages over longer distances UIC suggests to re-do the calculations for a single MFCN base station (varying the distance from the railway tracks from several meters to several kilometers), and using the above considerations. 9 Antenna factor In the draft FM54 report, for the calculation of Table 3: Resulting distance between MFCN BS and railway tracks, an 11dB antenna factor has been used to calculate the signal level received at the GSM-R antenna. However, UIC cannot understand this number. The reduction in antenna gain due to the receive antenna being outside the main lobe of a transmit antenna depends on both the antenna characteristics, the antenna heights and their mutual distance. The sample calculations below, based on Kathrein antennas, suggest that only high gain antennas O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 7 of 9
closer than some 70m from the rail track would have an antenna gain reduction of more than 3dB. In other words, the 11dB currently used in the FM54 report appear to be incorrect. Htrain antenna HTx Heff D start value step 4 m 20 m 16 m 10 m 10 m D m 10 20 30 40 50 60 70 80 90 100 fire angle degrees 58,0 38,7 28,1 21,8 17,7 14,9 12,9 11,3 10,1 9,1 Kathrein antenna type 80010642 G=18dB 3dB down at 13 degrees 742226V01 G=12dB 3dB down at 30 degrees Yellow means train antenna is within the -3dB points of MFCN transmit antenna lobe. O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 8 of 9
10 References Ref Name Source Date 1 EIRENE SRS 15.4.0 UIC 5 November 2013 2 3GPP 45.005 3GPP 3 MEASUREMENT REPORT Bundesnetzgentur 23/09/2013 COMPATIBILITY MEASUREMENTS GSM/UMTS/LTE VS. GSM-R Version 3.0 4 3GPP TR 43.030 3GPP 5 3GPP TR 45.050 3GPP 6 PT1(12)084/SE7(12)061 ECC PT1 September 21 2012 7 UMTS 900 vs GSM-R Study draft Telefonica Germany 13. June 2014 0.7.pdf 8 Guidance for improving coexistence between GSM-R and MFCN - Draft ECC / FM54 July 2014 O-8751-1.0_UIC_iput to FM54 on OOB 30-10-2014 page 9 of 9