Maritime Coverage in Dual GSM 900 DCS 1800 Network of big capacity. Planning, Integration and Optimisation

“Maritime Coverage in Dual GSM 900 DCS 1800 Network of big capacity. Planning, Integration and Optimisation.” Miguel Arcas Oliver, Juan José Flores Me...
0 downloads 1 Views 244KB Size
“Maritime Coverage in Dual GSM 900 DCS 1800 Network of big capacity. Planning, Integration and Optimisation.” Miguel Arcas Oliver, Juan José Flores Mederos Department Radioplannig. Telefónica Móviles España Jose Mª Hernando Rábanos, Rafael Herradón Díez, Florentino Jiménez Muñoz Department S.S.R..Universidad Politécnica de Madrid SPAIN.

Abstract: The present article explains and describes the selection of the empirical propagation model proposed in ITU recommendation P.370-7 [1], versus other models, for sea path propagation modelling, being based on measurements. The objective is the Planning and subsequent Optimisation of commercial sea lines in internal routes of the Balearic Islands Archipelago and of the Canary Islands Archipelago, as well as routes between the Balearic Islands and the Spanish mainland. The prediction models are refined with the new ITU model 1546/2002 [4], proposing its inclusion and incorporation in software tools for Planning and Optimisation. Finally, for Radio Planning, the emplacement characteristics and configuration of the Base Stations are described to provide maritime service with its aspects of Network Integration and Operational Optimisation. Key-Words: - Maritime coverage extended range GSM sea propagation

1. Introduction. Until the present time, the classic model has been used employing free space loss in the Planning, coverage and integration of Base Stations to give maritime coverage. The purpose of this study was to establish the correlation of real measurements with respect to current methods of calculating sea path losses and propagation over the sea. These tests were carried out in real sea lines in internal maritime routes in the Balearic Archipelago, the Canary Islands and on routes between the Balearic Islands and the Spanish mainland. For taking the measurements, the measuring set of Telefónica Móviles España (TME) was used, termed the SAM kit, which is composed of: Engineering Telephone Sagem 0T75-M dual and GPS, both with external antenna and with data output for PC serial port; a laptop PC gathered the measurements in a database. The result obtained showed a great similarity between the data provided by the model finally chosen and the real data. Some additional considerations are necessary for Planning maritime cover, sometimes in direct contrast with classic cellular concepts.

2. Study of the propagation model. Comparison with real measurements. Transmitter (Base Station) and receiver (Mobile Station) are characterized, allowance is made for

position, transmitter powers, radiation patterns, antenna directions and losses in order to have the real equivalent isotropic radiated power, EIRP, for comparison with the models.

2.1 Compared models: The empiric model based on ITU recommendation P.370-7 [1]: This model is applicable in any case, without any angular limitation. Tabulated data based on extensive measurement campaigns. A transmitter is assumed of 1 kW effective radiated power. E.R.P. There are tables that provide the Electric Field value exceeded in certain percentages of emplacements and time, produced as a function of transmitter altitude, transmitted frequency, the distance and the type of terrain (warm sea for our case). Some default values are assumed and some additional formulas and graphs for the deviations in the percentage of emplacements, undulating nature of the terrain and obstructions near the reception point. The corrections are also given for different altitudes of Transmitter and Receiver. The reflection pattern described by the UPV [2]: This is a theoretical model, although it admits experimental values obtained from the analysis of marine activity, to characterize roughness due to wave action, which produces a dispersion effect. It should be guaranteed that the angle of the hypothetical ray reflected on the sea exceeds a minimum value given by: Ølim (mrad) = (5400/f(MHz) )1/3






The fundamental ray is direct. Allowance has been made for the changing nature of the sea, the continuous movement of the waves, roughness. Thus, when a ray impacts on the surface of the water, besides the mirror reflection, myriad rays arise in different directions, constituting the incoherent component of the field. This model is applicable provided the Base Station is sufficiently near the coast, thereby assuring that the reflection that takes place is on the sea and that the constants used in determining the reflection coefficient are those of the sea.

-50,0 -60,0 -70,0 -80,0 -90,0 -100,0 -110,0 d











2.2. Values measured: The data used in the study have always corresponded to real GSM calls with their corresponding handover between cells. For this reason the data have been processed in many short hops by the inherent evolution of the call. In this way a great collection of data has been made available, many of them momentary in nature. In figure 1 the result provided by the models can be seen, together with the basic free space losses and the measured value. The measured value is taken as reference and the % represented on it.

(Figure 2)

In it, a general characteristic can be observed in long distance measurements with dominant servers: the signal continues fairly true to the model P.370-7 for 50% of the emplacements and 50% of the time, which in turn comes close to the free space loss. The UPV model provides a value less than that measured. The second representative lane, in figure 3, corresponds to a mobile telephone on the deck of a small pleasure boat in the vicinity of the coast, at 3 to 8 km: -35




That of diffraction for spherical earth according to ITU recommendation P.526-E [3]: The spherical earth model is applicable only for links over the sea and does not contemplate dry land characteristics. This is justified because in the earth’s curvature it produces an obstruction of the first Fresnel zone associated with the path, whereby the diffraction phenomenon occurs. This model will be applied when the diffraction model condition is not satisfied, minimum value given by (1).

P-370 -55


-65 -75 -85 -95 d














30 20

(Figure 3)

10 0 -10 -20 -30 -40 -50 -60 0









UPV P-370-7 FREE 80 UPV P-370-7 FREE

(Figure 1)

Also provided as examples, are two of the lines measured, considering them to be representative, showing the measured value preceded by that obtained from the models: The first of them, in figure 2, corresponds to the service provided by a dominant station to a mobile telephone on the deck of a ferry at a distance of 30 to 75 km away:

In this case a general characteristic can be observed for short and medium distance measurements and stations not characterized as dominant: a great resemblance is observed between the ITU model and that of UPV, the free space losses differing.

2.3 Results obtained: As a first result, it can be mentioned that the angular condition that characterizes the scope of application of the diffraction model has not been fulfilled in the measurements carried out and therefore there are no comparisons for this model. The foregoing is logical because the altitude of the stations intended to give service at sea is considerable and so the condition of grazing incidence is not fulfilled. The ITU-P-370 model corresponded at all times to the measurement. As the altitude and distance of the server station increase, there is a greater predominance of the line-

of-sight effect, since the ITU-P-370 model and the measurement tend to the free space losses. At medium and short distances and low server station altitudes, it seems to adapt to the UPV model. The conclusion reached is that the P.370-7 model is that which in all cases is most true to the measured result.

3. Network Planning of base Station for maritime coverage. 3.1 General problem: The first thing that should be kept in mind is that it is a question of giving coverage at sea to an extensive GSM network. For this reason it is not a matter of a simple connection characterized by its S/N ratio, but rather it is also necessary to consider the classic problems of integration of a cellular network taken to the extreme. Here the classic cellular network concepts of cell boundaries as demarcated as possible, come into conflict with the need for long distance propagation. Obviously, for reasons of synchronisation, the extended range facility [6] will be used that allows enlargement of the service distance of a GSM cell, from 35 km up to 121 km. Thus, in the ideal case, it is possible to coverage up to 242 km of continuous sea lane from both ends of the route.

3.2 Predicting the propagation. As has been stated, the basis for the propagation model proposed is in the recommendation: ITU-R P-370-7 [3]. The prediction tools consider the visibility of the Base Stations with respect to the sea, there being a clearance toward the latter determined by the radius of the first Fresnel zone. This should be kept in mind for all Base Stations, even when these do not have the sea as their objective, in order to facilitate the calculation of interference for a determined frequency plan, in order to allow for the undesired effect of overreach that any station can have toward the sea. For implementation purposes, the model mentioned has been employed in its new version: ITU-R 1546/2002. [4] for its greater generality and including validity in new systems as UMTS.

The precise data are given next, in figure 4, the model provides the field for 1 kW E.R.P. or the basic propagation losses for a receiver at a standard reference altitude. The validity of the model is for distances > 1 km, and frequencies of 30-3000 MHz. Correction is applied for a receiver altitude different from the reference altitude, with validity for receiver altitude > 3 m. Thus it is possible to determine the level expected at each point, the best server cell and, once the frequency plan is introduced, the C/I ratio. It must be pointed out in this last point, and given that the model allows it, that the expected field level can be estimated for the best server, as that which the model provides for a certain t (%) of time (typically 50%); however when evaluating the interfering field, one can be pessimistic and apply a value for a smaller percentage time.

3.3 Determining Base Station emplacements to give maritime coverage. When defining the candidate stations to give long distance maritime cover, one usually encounters a mature GSM network, with normal MACRO stations and UMBRELLA stations usually located in dominant points. The emplacements where UMBRELLA Base Stations are installed are the natural candidates for installing a new EXTENDED RANGE Base Station. Although there exists the possibility that a normal UMBRELLA station may also implement extended range cover, the most convenient solution (justified below) is to install a new EXTENDED RANGE Base Station in the same location as the UMBRELLA Base Station. It is not necessary to fit extended range in all the emplacements where the UMBRELLA facility is present, a study of the terrain and the network will determine where.

3.4. Reasons for dedicating a Base Station exclusively for extended range service: The UMBRELLA station has its antennas inclined with a vertical beam width that does not extend beyond the 35 km typical of GSM, with the consequent loss for vertical misalignment. See figure 5.




Time (%)




(300-3000 Mhz)


h.Tx over terrain

(meter >10)





(Km < 121 Cover.GSM)


Rx height

(meter, > 3m.)

h.Terrain over sea


Extended Range Km. 160121Km


35 Km.

25 Km

(Figure 4) (Figure 5)

The orientations of the antennas need not coincide with the targets of the long distance station, with the consequent loss for horizontal misalignment. The EXTENDED RANGE Base Stations have a configuration different to that of the UMBRELLA Base Stations, in line with the long distance objectives, and in this way interference is reduced between the EXTENDED RANGE, UMBRELLA and MACRO service areas. Since two time slots have to be reserved for each channel, it is possible to make the number of carriers unviable in the case of implementation in the same UMBRELLA station. As the cells are different to those of normal range, this simplifies their Control, Integration and Optimisation, especially for traffic control to reduce interference. Since the objective of the extended range station is usually on the horizon, the antennas can be located low (less cable attenuation), they can be bigger (special antennas like parabolas, parabolic section antennas), and be supported by the framework of the tower. This point, on some occasions, can be very good for avoiding interferences, as can be seen in figure 6. Extended range GSM antenna cannot see UMBRELLA and MACRO area services, caused by its low location, own geometry of emplacement makes the rest.

It is also necessary to identify the MACRO stations that will give maritime service, inshore, to adjust their configurations. It is convenient to take the stations nearest the sea, relatively high (> 35 m) and positioned so that their horizontal beam can be pointed at the sea without overreaching undesired areas, on the other side of a bay with a high station density for example, as is shown schematically in figure 7. The antennas that serve toward the sea should have a slight inclination, < ½ vertical beam.

(Figure 7)

The basic idea is to get the MACRO stations to take all the traffic possible from the UMBRELLA Base Stations, and these in turn do so with regard to the EXTENDED RANGE Base Stations, for the purpose of reducing interference in upper layers.

3.6 Stratified schematic of coverage. Extended Range 160 Km


25 Km (Figure 6)

For configuration, the antennas chosen will have the largest possible gain, with the smallest vertical and horizontal beam widths that allow the objectives to be reached, having the requirements already described and with the considerations envisioned in figures 5 and 6, with regard to the vertical widths, the inclinations and the positions of the antennas. The use of low noise level preamplifiers is considered indispensable in the in reception tower, as well as space diversity in reception.

3.5 Considerations in stations to serve in near close maritime coverage. It is convenient to readjust the configurations of the UMBRELLA stations in order to adapt the transition of coverage with the extended range stations, generally by altering the inclination or changing the antenna for another vertical beam value.

We have a GSM–900 cellular network that from the point of view of coverage layers, and from its lower layer, consists of MACRO, UMBRELLA and EXTENDED RANGE stations that have to be integrated with a similar network that is across a stretch of sea on another island, peninsula or arm of a bay. The layer structure is determined and the system works with synthesised hopping frequency [7] plans with total synchronisation of UMBRELLA and EXTENDED RANGE Base Stations at the same emplacement and in random mode with regard to the remaining stations. There are two bands: GSM-900 and DCS-1800 with more frequencies in the DCS-1800 band (although with less cover). To enlarge the band where working with random use of frequencies, in order to enhance overall network quality, by reducing interference, twinning is planned of DCS-1800 stations with those of GSM-900, hence the network reads from the lowest layer: MACRO DCS, MACRO GSM, UMBRELLA DCS, UMBRELLA GSM, EXTENDED RANGE DCS, EXTENDED RANGE GSM. The appropriate layer administration software together with all Base Station considerations and Integration and Optimisation must assure correct and integrated operation in the maritime coverage network.

4. Integration and Optimisation in Base Station Network for Maritime Coverage. 4.1 Boundaries neighbour. The EXTENDED RANGE Base Stations are treated as one more layer in the layered hierarchy of the network, with some specific distinctions for each manufacturer and a few adjustments. Lowering the level will be done after a certain period of monitoring a good neighbour level. Raising the level must be carried out quickly, in the event of any abrupt fall in level or quality. To descend from a given GSM or DCS level, one only drops to the GSM level immediately below, (for example from EXTENDED RANGE DCS or EXTENDED RANGE GSM, one will always descend only to UMBRELLA GSM and never to UMBRELLA DCS). There are two reasons for this: it is necessary to limit the number of boundaries and the descent always has more probability of success on a GSM sector parallel to another of DCS than on to DCS due to the better propagation and fading conditions of GSM-900. Furthermore, there is no impediment that the destination GSM repeats its descent onto its parallel DCS. To ascend a level, the same approach is used, one only goes up to the GSM layer. The foregoing remains valid even though all the levels are not present, two-level jumps are also possible, since they can be necessary in certain cases. All the above-mentioned is shown schematically in figure 8.

necessary to control possible ping-pong effects due to a layer handover being tripped after one for distance. Such an occurrence has to be detected by means of OMC tools, verification measurements in important sea lanes, etc. It would be corrected by adjustment of the layer parameters, distance handover parameters, addition or suppression of boundaries, adjustment of radiation patterns, range definition of extended range facilities in emplacements that do not have them, etc.

4.3 Frequency Plan. The frequency plan is based on distributing one part of the band for BCCH carriers and the rest for synthesized hopping [7] traffic carriers in 3/1 reuse for the MACRO layer. The adopted solution has been to reserve some BCCH frequencies for the extended range cells (which shall only be reused with extreme care in highly demarcated areas). The solution adopted for both the UMBRELLA and the EXTENDED RANGE cells, is to assign all the frequencies available for hopping in 1/1 reuse. It is very important that all the sectors in a given location, in the same band, belong to the same logical site in order to maintain synchronism and in this way, interference never takes place, not even in adjacent channel, between any time slots of any traffic carrier of the emplacement. The foregoing is valid for a number of traffic carriers the same as or smaller than half the frequencies available in the hopping sequence. All that explained in the frequency plan is repeated in identical fashion in each of the bands.

5. Conclusions.

Extended range GSM

Faced with a complex problem, a valid, simple and programmable model has been proven in the field for predicting the field that a GSM-900 or DCS-1800 (with validity in UMTS band) transmitter will produce for maritime coverage. The basic considerations have been given regarding Planning and Integration in a network of stations for maritime coverage and the features to vary in the network for the adequate operation thereof.

Extended range DCS Umbrella GSM

Umbrella DCS

Macro GSM

Macro DCS

(Figure 8)

4.2 Handover by distance. The handover by distance should be activated for all cells at a value the same as or less than the range of the cell, 35 km for normal cells or 121 km for those of extended range. A reasonable approach is to define a value between 30 and 35 km in the UMBRELLA cells and a smaller value in the MACRO cells. It is

References: [1] Recommendation ITU-R P.370-7. [2] Rubio L., Rodrigo V.M., Juan-Llacer L., Cardona N., “Modelo de propagación para entornos marítimos”. UPV.URSI-SPAIN 1999 [3] Recommendation ITU-R 526 [4] Recommendation ITU-R 1546/2002. [5] Hernándo Rábanos J. M., Transmisión por radio, Editorial Centro de Estudios Ramón Areces S.A. 1995 [6] Luengo R., Yuste M. “Plannig and Optimization of Marítim Coverage”. Journal ITU-R. Madrid 1999 [7] Yuste M., Luengo R. “Introducción al Salto en frecuencias. “. Journal Telecom I+D 1996

Suggest Documents