Design of Mobile Cellular Coverage in Tunnel Environments

Transcription

1 Design of Mobile Cellular Coverage in Tunnel Environments Ramón Mª Ruiz Tarrés*, Florentino Jiménez Muñoz**; Rafael Herradón Díez**, José Mª Hernando Rábanos***. *TELEFÓNICA MÓVILES ESPAÑA, S.A. **DIAC-E.U.I.T. de Telecomunicación(UPM) Ctra de Valencia, Km. 7, MADRID. Tel: , ***SSR-E.T.S.I.Telecomunicación(UPM) ABSTRACT When planning cellular coverage in special areas, as tunnels or indoors, the continuity of connections in progress in entrances and exits is a critical point. A wider overlapping concept is introduced as a method to include environment relevant elements, like existing network or the mobile channel, necessary in order to achieve a correct design. As practical example the analysis in tunnels of a high-speed railway line is presented, where a dedicated dual GSM/UMTS coverage system is designed. Also a prediction model is proposed and utilisation of radiating cable instead of antennas is discussed the necessary overlapping for making possible that all handover processes can be done. We will also see that the necessary level it s not fixed only by the cells signal level involved in the handover and that it does not exist a symmetry between both overlap directions. II. HANDOVER TIMMING We will suppose that in instant t 1 the access takes place to cell A. Several timers begins at this time that avoid that the mobile station leaves the cell A during a defined period of time, depending on the conditions of the radio link. I. INTRODUCTION When a cell to cell handover should be done at the entrance or the exit, coverage overlapping between the cells involved must be ensured in order the handover process is correctly achieved. This necessary overlapping is usually defined like a distance in both directions, in which the signal levels from de cell A and B are high enough to make possible the handover. t1 Access to cell A HO Temporizations... ( 3-5 s ) ( 1-2 s ) t2 Start reports? t3 Cell B in reports ( 1-5 s ) t4 Mobile synchronized to cell B time Fig. 2. Time in HO process Detection A B Fig. 1. Handover process This overlap concept does not take into account all the characteristics of the environment and the involved cellular mobile system. This can lead to designs that, once done, do not achieve the desired service quality requirements. Modifying the installed systems is usually very expensive. As we will see, it s much more reasonable defining overlapping as the handover overlapping which means A B Independently of these timers, in time t 2 the mobile has received enough information from the network to begin to send information about neighbouring cells. In GSM the neighbouring information is send to the BTS each 480 ms using the associated channel SACCH. In UMTS this associated channel doesn't exist. The transmission of neighbour measurements is carried out when they occur certain "events" defined in specifications [1]. UMTS supports handover between different carriers and systems that can be classified by the way of operation of the system and for the execution mechanism [2]. According to the way of operation we can classify them in the following way: - Intra-mode Handover: Between two FDD or TDD carriers with the same or different frequency. - Inter-mode Handover: Between FDD and TDD mode.

2 - Inter-system Handover: Between different systems 3G or 3G-2G. Attending to the way of execution it can be classified as: - Hard Handover: Between different frequency carriers. The commutation to the new channel means liberation of the old one. There is no temporary overlapping among both connections. - Soft Handover: The mobile station establishes simultaneous connections through several stations. For this reason the communication is not interrupted during the transfer, contrary to the previous case. - Softer Handover: They are soft handovers between sectors of the same base station. The difference with soft handover is in the combination mechanism used in the uplink connection. In this document we refer to the intra-mode handover in FDD between different sites, at the same frequency (soft) or among different frequencies (hard). Soft handover provides an additional gain in front of fast fading, reducing the E b /N o necessary to assure the desired quality. usually between 1 and 5 seconds. However, it can be longer, depending mainly of the terminal manufacturer and the conditions of the radio link. Also the UMTS mobile should be synchronised to the neighbour before performing the handover. UMTS is an asynchronous CDMA system, so the cell search procedure o synchronisation procedure differs greatly from the procedure in a synchronous one like IS-95. In UMTS cells use different scrambling codes instead of different code phase shifts, nowadays terminal technology cannot search for 512 code of 10ms without any prior knowledge [3]. UMTS cell synchronisation procedure has basically three steps. Specification defines the process in a general way [1], but there are no requirements as to which steps to perform and when. The practical implementation is open to the manufacturer. Synchronisation properties scheme need to be taken into account when setting network parameters. For handover optimal performance, target cell search in connection must be optimised. For example, correct planning of scrambling codes can improve it. In t 3 the mobile detects cell B. Time between instant t 3 and t 2 is very variable depending on the following factors: Cell B level and mobile channel conditions. Cells in service with same or better level that B. Candidate set of neighbouring cells in UMTS or ACTIVE cell list in GSM. Multi Band Cells Reported parameter (MBCR) in dual mobiles GSM/DCS. Mobile stations report information about neighbouring cells. In GSM mobile reports include information on the received level and the BSIC of up to 6 neighbouring cells. As much as higher is the number of neighbours more cells they will be excluded of the report. New neighbours in service can retard the entrance in the report of cell B and, consequently, a handover to a tunnel or interior fails where it was perfectly adjusted before. In the same way, the MBCR parameter determines the number of neighbours reported by each band in dual mobiles. Their modification affects at the moment that our cell B enters in the reports and, therefore, affects to the handover adjustment. Once the terminal has detected cell B, it needs to obtain the synchronisation with the cell. In GSM, the mobile must read the channel SCH of the neighbouring cell. Theoretically, due to the GSM temporary scheme, a mobile terminal connected to a TCH/FR channel needs between 0.6 and 2.28 seconds (average value=1.38s) for synchronising to a neighbour and send the first report to the system containing this neighbour. In practice, tests carried out with several mobiles and real traffic analysis shown that this time is t4 Mobile synchronized ( 1-3 s ) to cell B t5 Enough reports? t6 HO Decision ( 0.6 s GSM ) Fig. 3. Time in HO process Execution t7 Mobile served by cell B time At the moment network receives the first report of a neighbouring cell (instant t 4 ) it is already possible to try to carry out a transfer toward it. However, the system generally evaluates several reports to assure that the handover is performed at the right moment. The network operator determines the number of reports. In t 5, radio network controller has enough number of reports to decide. Handover algorithms will determine the moment when handover should be performed (t 6 ). Handover algorithms are also opened to the supplier. Once the handover decision has been taken, the system and the mobile begin the process that will take the mobile station to be served by cell B (t 7 ). This process takes about 0,6 seconds in GSM under normal quality conditions. Summarising, to evaluate the handover overlapping first of all it is necessary to determine the geographical point

3 where wanted neighbouring cell B is detected as a possible candidate. Starting from this time (t 3 ) it is possible to evaluate the necessary overlapping time using equation (1) and (2). = ti + 1 ti (1) ti 6 t ho3 = = ti (2) In GSM system their minimum value is: 6 tmho3 = ti = = = 7. 04s (3) i 3 III. HANDOVER IN TUNNELS We are going to examine the specific problem with handover in tunnels. We will suppose a tunnel with only one indoor cell (I) that assures the continuity of the calls inside. i 3 There is only one handover origin cell (cell I) and several exit cells candidates. It is enough to assure the overlapping time with anyone of them (expansion effect). Before doing the handover, the only existing cell is the server cell (cell I). Therefore, the exit cell enters in the neighbouring report when he is strong enough to be detected. The average level necessary for the exit cell to be detected will be: Lmint 3 = S + AtV + AtC + Mg (4) Where: S: Mobile Sensibility AtV: Vehicular loss AtC: Body loss Mg : Fading margins C. Tunnel inside handover Células de entrada E Célula interior I Células de salida S In tunnel inside handover case, we can suppose that there are only two cells involved (I1 and I2). The minimum overlapping time will be t mho3 seconds since I2 cell average level is better than Lmint 3. As in the previous case, cell I1 should maintain a better level than Lmint 3 during this whole time. Therefore, the existing overlapping time is the time when the signal levels of both cells are better than Lmint 3. IV. HIGH SPEED RAILWAY TUNNELS A. Tunnel Propagation Model Fig.4. In HO (concentration effect) and Out HO (expansion effect) A. In Handover The entry cell or E cells are the set of possible servers before entering in the tunnel. It is usually the most problematic handover for the following reasons: It is necessary to assure overlapping handover with cell I for each one of the possible entrance cells. The mobile is in a multiple neighbouring cells environment that will retard the entrance of cell I in the reported list. The real problem is that cell I enters in the neighbouring report (instant t 3 ) when cell E still has t mho3 seconds in those the cell E radio link has enough level and quality to assure the whole handover process. B. Out Handover Tunnels in high-speed railway lines are straight-line tunnels or their curvatures have a very wide radius. Their transversal section is large due to tunnels should allow both circulation lines. In addition, a large section decreases the pressure wave when trains enter in the tunnel. The pressure wave effect is also the reason to reduce the number and size of structures inside the tunnel, as the on-fly power supply system based on aerial wires. In these environments, a coverage tunnel system based on antennas show a guide-wave effect and no blocking effect when the LOS area is left. Besides, the train blocking losses is low because the train has a smaller transversal section than the tunnel. Figure 5 and 6 shows the average signal strength curves and the model curves for systems working at 910MHz and 2GHz in a typical high-speed railway tunnel, obtained in a measurement campaign in a Spanish highspeed railway line. The cell or exit cells S are the possible candidates to receive the call from the indoor cell I. In this case, the out handover benefits of the following:

4 called breaking point, the loss slope changes. In some situations the breaking point could be the fresnel breaking point. In a rectangular tunnel and transmission and reception antennas are placed on the middle of the tunnel, the distance up to the breaking point is: 2 2 d Minimum width, height (7) λ λ In practice, the breaking point can be fixed evaluating the slope changes in the trial signal strength curves. This method is appropriate for when they are no placed in the middle of the tunnel. Fig. 5. Averaged received signal and model at 910 MHz The average signal strength curve is the result of averaging all the measures taken inside the tunnel, changing the transmission antenna position and the receiving paths. Near the antenna, the real antenna diagram pattern causes the difference between the average signal strength curve and the model curve because the last uses the main features of the diagram pattern, but not the complete diagram. Non-Line-Of- Sight (NLOS) section In the NLOS zone the blocking and slope change model is used [5]. l log( d + 10 n1 + L 1 rup ) + [ log( dlos / drup) ] + 10 n2 log( d / d LOS + ) d > d LOS (8) L 1 is the blocking losses between the LOS zone and the NLOS zone. As mentioned before, this blocking attenuation can be fixed to zero in high-speed railway tunnels. The figure 7 shows the model parameters and the error model curve. Fig.7. Model parameters B. Penetration loss. Power Budgets Fig. 6. Averaged received signal and model at 2 GHz The model curve responds to: Line-Of- Sight (LOS) section In this area the hybrid model [4] is used. It assumes two regions in the LOS section. l log( d) l log( d ) + 10 n [ log( d / )] drup < d < dlos rup 1 drup Where l o is the free space loss at 1 m. d < d rup (5) l 0( db) = 27, log( f ( MHz)) (6) In the nearest transmitter zone the model use the freespace propagation formulas. If the distance is above the An important issue planning railway coverage is the high attenuation that its boxcars present to the inside propagation. In the case of high-speed lines, in some models the windows are notably small and sometimes their glasses have been covered with metallic materials, increasing still more the radio-electrical attenuation. Telefónica Móviles España reports, about penetration measurements in some train models, have studied the effect of the incidence angle and the materials. Results show that incidence angle less than 5 respect of the railway line can bear losses up to 38dB for GSM and 30dB for UMTS. With normal incident angle (90 ) losses are less than 12dB for GSM and 9dB for UMTS. It is important to highlight that using antennas, the incidence angle is very low in almost all the way and near 0 in the cell coverage limits. However, in case of using radiating cable we can consider that the incidence angle is 90 so the penetration losses are minimum. Another radiating cable advantage is the non-existing

5 Doppler effect. For all these reasons we should take into account the radiating cable as a good alternative to cover high-speed railway tunnels, even for high frequencies as UMTS and in spite of the high cost of the cable and their installation. Figure 8 shows a compensating attenuation comparative table for GSM and UMTS, using antennas or radiating cable. The Rayleigh margin used for this environment is 4 db [6]. Fig. 10. Attenuation in a high-speed railway tunnel V. CONCLUSIONS Fig. 8. GSM and UMTS Power Budget in a tunnel Figure 9 shows the maximum length that can be reached for two types of Andrew s radiating cable [7]. A wider coverage-overlapping concept has been presented. The effect of existing cells, mobile channel features, network parameters or cell synchronisation procedures over the radio network performance and how the handover overlapping concept allows include these factor in the whole design has been shown. GSM and UMTS handover timing keys are explained. The necessary handover overlapping coverage in several situations is analysed and calculated. Also a model for high-speed tunnel propagation prediction and utilisation of radiating cable instead of antennas is evaluated. Finally, an example of application of the concepts introduced in this paper is shown. Fig. 9. Maximum length for GSM and UMTS Figure 10 shows the predicted propagation attenuation for GSM, using the proposed model, in a tunnel of 1510m, with two antennas in the tunnel ends. There is not direct vision among both ends. It can be observed that, with the proposed solution, only an overlapping of approximately 400m is obtained. It can seem enough. However, the handover overlapping for this case, when a handover inside the tunnel must be performed, it should be of about 685 m for a maximum speed (350 Km/h). A possible solution to get the handover overlapping can be to modify the position of the antennas or to use radiating cable instead of antennas. REFERENCES [1] 3GPP Technical Specification and , [2] José Mº Hernando Rábanos, Cayetano Lluch Mesquida, Comunicaciones Móviles de Tercera Generación, Telefónica Móviles España S.A., 2000, Tomo-I, Parte 2. [3] Harri Holma and Antti Toskala, WCDMA for UMTS, Ch. 6 Physical Layer, [4] Y.P.Zhang, A Hybrid Model for Propagation Loss Prediction in Tunnels, Proc-Millennium Conference on Antennas and Propagation, Switzerland, 9-14 April [5] Universidad Politécnica de Madrid - Grupo de Radiocomunicaciones Móviles, "Caracterización de instalaciones para comunicaciones móviles", Oct-1999, Informe 3. [6] Farrokh Abrishamkar, James Irvine, Comparison of Current Solutions for the Provision of Voice Services to Passengers on High Speed Trains, IEEE - VTC 2000, pp [7] ANDREW Series RADIAX - Radiating Cable,