GSM In Train Pathloss

Key words : in-train, penetration loss, propagation___________________________________________________

In-Train Penetration Loss and Design Recommendations for UMTS Network Deployment

Dr. Hatem MOKHTARI, December 2002

Author : Dr. Hatem MOKHTARI____________________________________________________ 1/5


1. Scope of work This document aims at providing a clear insight into in-train penetration loss to be used for design purposes. It also encompasses case studies performed within live Network Operators, along with a brief theoretical description of the propagation issues related to the upper UHF frequency band. More importantly, this document will be used as a reliable reference for RANN/Ben/Dutchtone UMTS Network deployment so that all the parties will use the same RF parameters, and comprehend the theoretical and practical reasons behind the adopted choice. This should avoid further misunderstanding and consolidates the effective collaboration between the parties. 2. Introduction Needless to say if the in-train penetration is important in a Cellular network Indeed, as the service is based on mobility and the subscribers are theoretically supposed to be anywhere, it is of great importance to consider the in-train penetration issues. Several studies have been conducted so far in the GSM900/GSM1800 bands but a few of them only in the UMTS band (around 2 GHz). In a microcellular environment, such as in street canyons and indoors the RF signal behaves similarly in both UMTS and GSM1800 bands. This is due to the fact that the higher the frequency the better is the propagation. We can easily explain this phenomenon by the fact the scattering characteristics of microwave signals tend to an asmptotic behavior. Reflection coefficient of semi-conductive materials tend to the unit, which explains the guided-wave behaviour in confined media (Tunnels, street canyons, airport terminals, etc.). However, the only characteristic that should make the difference between the two systems is the bandwidth when modulation is considered, but when pure RF carriers are transmitted GSM1800 and UMTS have the same characteristics. This document provides the propagation assumptions made to cope with the wider range of cases that might be encountered in a live UMTS network. Namely, grazing incidence (near 0 degree) and perpendicular incidence (near 90 degrees) will be explained and statistical results, performed on a real live network, will be presented. 3. Problem Statement Figure 1 depicts the basic assumptions made for the sake of in-train penetration study :



Train windows

Transmitting Node B Antenna (assumed as a Huyghens source)

Grazing incidence line (dashed line)

Incidence angle is a random variable wich is uniformly distributed between 0 and 180 degrees

When the trains moves within a cell coverage area the propagation is proved to be in grazing incidence, except for a few range of random angles. Therefore, the penetration loss is maximum. In a tunnel, however, in case where leaky-feeders are deployed, the propagation is under 90 degrees, which gives better in-train coverage. Let us now take account of the the angular spread combined to the loss through windows. In many cases, and especially in modern fast-speed trains, the glass is mixed with a semi-metallic material. This obviously leads to a more reflected energy and therefore less transmitted energy within the train wagon, which is not the desired situation for in-train coverage. Low in-train loss Zone

High in-train loss zone (very low incidence angles)

High in-train loss zone (very low incidence angles)

For our design purposes worst-case senarios are considered. In other words, the penetration loss values are taken from grazing incidence and not from perpendicular incidence because the latter leads to under-dimensioning the number of sites required. 4. Results 4.1. Studies in France



INRETS, a French institution made some measurement campaigns in-train for the French fast speed train (TGV) and found out that : Penetration Loss between 25 and 30 dB for grazing low-angle incidence, and between 2 to 5 dB only at 90 degrees incidence for DCS1800 frequency band. The result is found to be very similar to UMTS as the frequency bands are quite close to each other. 4.2. Studies in Spain Telefonica Moviles found similar results as for UMTS a penetration loss of 30 dB using antennas in tunnels. However, they recommend the use of leaky-feeders as the loss was found to be 15 dB. The leaky feeder was recommended because the propagation is perpendicular to the axis of the train, which therefore implies much less penetration loss as confirmed in the Frech case. Furthermore, the Doppler effect is non-existant in case of leaky-feeder solution. The drawback is, obviously, the cost of such a solution. Also, we can see from the table below that there is no correlation between the penetration loss and the bit rate, which is an important point to mention as it will ease our design by avoiding too much parameters for each service (speech or data).

4.3. COST231 assessment of in-train losses COST231 mentioned that the in-train penetration losses are similar to those found by the Spanish and the French studies. Values between 25 and 30 dB for grazing incidence are found. However, the COST231 study (made in TGV Nord ParisLille) mentioned a very interesting issue regarding synchronisation in GSM. The figure below shows how synchronisation problems arise (i.e. no neighbours can be



reported) when the train speed exceeds 200 km/h. In UMTS we do not have similar studies but this problem might occur and we need to mention it.

5. Conclusions and Recommendations According to the abovementioned studies, UMTS radio related characteristics are very similar to DCS1800 ones except bandwidth. Therefore, we recommend to use 25 dB in-train penetration loss to be added in our linkbudget calculations for high-speed trains with semi-metallic windows, and 15 dB for classical trains.. We also need to avoid the use of 5 dB loss because this value has been used for tunnels where leaky-feeders were installed. In fact, leaky-feeders have the best propagation characteristics in a confined environment as the field strength, from one hand is uniformly distributed and from the other hand has an incidence angle of 90 degrees, which is unlikely to happen outdoors from outdoor cells using antennas instead. 6. Technical References [1] U. Dersh, J. Troger, E. Zollinger, “ Multiple Reflections of Radio Waves ina a Corridor”, IEEE Trans. On Antennas and Propagation, Vol. 42, no. 11, Nov. 1994. [2] J.-F. Sante, “Measurements and Modelling in Rural Areas”, COST231 TD(93) 121, Limerick, Ireland, September 1993. [3] F. Ikegami, S. Yoshida, M. Umehira, “Propagation Factors Controlling Field Strenght on rban Streets”, IEEE Trans. On Antennas and Propagation, Vol.32, No. 8, August 1984, pp. 822-829.



[4] COST231, “Urban Transmission Loss Models for Mobile Radio in the 900 and 1800 MHz Bands (Revision 2)”, COST231 TD(900 119 Rev. 2, The Hague, The Netherlands, September 1991. [5] R. M. Ruis Tares, F. J. Munos, R. H. Diez, J. M. H. Rabanos, “Design of Mobile Cellular Coverage inTunnel Environments”, Telefonica Moviles Espana Report.


GSM In Train Pathloss

Recommend Documents