Radio Planning Considerations in TETRA to LTE Migration for PPDR Systems : A Radioelectric Coverage Case Study

Transitioning a Terrestrial Trunked Radio (TETRA) network to a Long-Term Evolution (LTE) network in public protection and disaster relief (PPDR) systems is a path to providing future services requiring high radio interface throughput and allowing broadband PPDR (BB-PPDR) radio communications. Users of TETRA networks are currently considering how to deploy a BB-PPDR network in the coming years. This study offers several radio planning considerations in TETRA to LTE migration for such networks. The conclusions are obtained from a case study in which both measurements and radioelectric coverage simulations were carried out for the real scenario of the Murcia Region, Spain, for both TETRA and LTE systems. The proposed considerations can help PPDR agencies efficiently estimate the cost of converting a TETRA network to an LTE network. Uniquely in this study, the total area is divided into geographical areas of interest that are defined as administrative divisions (region, municipal areas, etc.). The analysis was carried out using a radio planning tool based on a geographic information system and the measurements have been used to tune the propagation models. According to the real scenario considered, the number of sites needed in the LTE network—for a specific quality of service (90% for the whole region and 85% for municipal areas)—is a factor of 2.4 higher than for TETRA network.


Introduction
The conversion of a Terrestrial Trunked Radio (TETRA) network to a Long Term Evolution (LTE) network in public protection and disaster relief (PPDR) systems is a possible solution to provide future services requiring high radio interface throughput and allowing broadband PPDR (BB-PPDR) radio communications [1].In Report 218, The Electronic Communication Committee (ECC) proposed a roadmap until 2025 for the transition to broadband communication in PPDR systems [2].This roadmap envisages coexisting TETRA and LTE networks for several years until the LTE system has all the functionalities of the PPDR systems.Therefore, the introduction of LTE for PPDR should complement, not replace, existing TETRA networks, which will continue to be the best choice for short-term mission-critical voice service.
In addition, PPDR users cannot leave their current TETRA systems until a new mobile broadband network is built that is able to provide radioelectric coverage equal to or better than the radioelectric coverage currently provided by TETRA systems.Therefore, before any broadband solution can replace the current TETRA systems, the LTE network should meet all radioelectric coverage requirements currently satisfied by the existing network.
The most plausible future scenarios to deliver the increasingly data-intensive applications demanded by PPDR agencies are expected to rely on the use of both dedicated and commercial LTE networks [3].For rural areas, the base stations of a radio communication system are located in the mountains, and the radioelectric coverage has 'dark zones' (without radioelectric coverage), usually in forests, mountains, rivers, etc.In these zones, operators are not interested in deploying new base stations to offer broadband communications.However, in urban environments, operators have already deployed commercial broadband networks, that could be used by PPDR agencies for broadband communications.Therefore, the possibility of a hybrid model (dedicated network for rural areas and commercial networks for urban areas) could be an interesting solution for BB-PPDR radio communications.
More than 114 countries around the world have deployed TETRA networks (at the regional or national level) in the past years according to The Critical Communications Association (TCCA) [4].Now, the users of these networks are considering how to proceed to achieve a BB-PPDR network in the coming years.Therefore, PPDR agencies are interested in analyses of the many issues (cost analysis of the resources needed, the expected radioelectric coverage, quality of service, etc.) involved in planning the transition of a TETRA network to an LTE network.
A framework for the modeling and planning of an LTE public safety broadband network has been presented by Rouil et al. [5].The study was carried out on a nationwide scale, dividing the total area into 20 × 20 km squares and establishing a minimum percentage of radioelectric coverage in each square.
In this study, several radio planning considerations in the TETRA to LTE migration for PPDR networks are offered.In this case, different from Rouil et al.'s work [5], the total area is divided into geographical areas of interest that are defined as administrative divisions (region, municipal areas, natural parks, etc.).To illustrate this proposal, a radioelectric coverage analysis in a rural area with an existing TETRA network that has to transition to an LTE system has been carried out in a real scenario by means of both simulations and measurements.The analysis was performed by taking advantage of the potential of a radio planning tool [6] based on a geographic information system (GIS) [7], and the measurements were performed and used to tune propagation models.The proposed methodology can help PPDR agencies efficiently estimate the cost of transitioning a TETRA network to an LTE network.

Scenario
RADIECARM is the TETRA network deployed by the Regional Government of Murcia (Spain) that is used by the emergency and safety services (fire brigade, police, forest police, etc.).All these services are coordinated by the Emergency Coordination Center 112.RADIECARM is a dedicated network with 16 base stations located in mountains and around 2000 terminals.In Figure 1, we can see the typical infrastructure (tower, equipment, etc.) at a site and, in Figure 2, a mobile terminal mounted in a four-wheel drive vehicle.Figure 3 shows the 16 TETRA base stations (red circles).
The most plausible future scenarios to deliver the increasingly data-intensive applications demanded by PPDR agencies are expected to rely on the use of both dedicated and commercial LTE networks [3].For rural areas, the base stations of a radio communication system are located in the mountains, and the radioelectric coverage has 'dark zones' (without radioelectric coverage), usually in forests, mountains, rivers, etc.In these zones, operators are not interested in deploying new base stations to offer broadband communications.However, in urban environments, operators have already deployed commercial broadband networks, that could be used by PPDR agencies for broadband communications.Therefore, the possibility of a hybrid model (dedicated network for rural areas and commercial networks for urban areas) could be an interesting solution for BB-PPDR radio communications.
More than 114 countries around the world have deployed TETRA networks (at the regional or national level) in the past years according to The Critical Communications Association (TCCA) [4].Now, the users of these networks are considering how to proceed to achieve a BB-PPDR network in the coming years.Therefore, PPDR agencies are interested in analyses of the many issues (cost analysis of the resources needed, the expected radioelectric coverage, quality of service, etc.) involved in planning the transition of a TETRA network to an LTE network.
A framework for the modeling and planning of an LTE public safety broadband network has been presented by Rouil et al. [5].The study was carried out on a nationwide scale, dividing the total area into 20 × 20 km squares and establishing a minimum percentage of radioelectric coverage in each square.
In this study, several radio planning considerations in the TETRA to LTE migration for PPDR networks are offered.In this case, different from Rouil et al.'s work [5], the total area is divided into geographical areas of interest that are defined as administrative divisions (region, municipal areas, natural parks, etc.).To illustrate this proposal, a radioelectric coverage analysis in a rural area with an existing TETRA network that has to transition to an LTE system has been carried out in a real scenario by means of both simulations and measurements.The analysis was performed by taking advantage of the potential of a radio planning tool [6] based on a geographic information system (GIS) [7], and the measurements were performed and used to tune propagation models.The proposed methodology can help PPDR agencies efficiently estimate the cost of transitioning a TETRA network to an LTE network.

Scenario
RADIECARM is the TETRA network deployed by the Regional Government of Murcia (Spain) that is used by the emergency and safety services (fire brigade, police, forest police, etc.).All these services are coordinated by the Emergency Coordination Center 112.RADIECARM is a dedicated network with 16 base stations located in mountains and around 2000 terminals.In Figure 1, we can see the typical infrastructure (tower, equipment, etc.) at a site and, in Figure 2, a mobile terminal mounted in a four-wheel drive vehicle.Figure 3 shows the 16 TETRA base stations (red circles).

Radio Planning Tool
To carry out the radioelectric coverage analysis, the application RADIOGIS [6], which is a tool developed by the authors for the management and calculation of the radioelectric coverage of radio communication systems such as GSM, UMTS, LTE, TETRA, TDT, WiFi, etc., was used.RADIOGIS works on Windows PCs and is integrated with the GIS software ArcGIS 9.1 from ESRI.RADIOGIS has, among others, the following functionalities: a) Calculations of power, electric field, or power density radioelectric coverage while being able to select the propagation model to be applied: ITUR-526 [8] and ITUR-1546 [9] for rural environments; Okumura-Hata [10], COST-231 [11] and Walfisch-Bertoni [12] for urban areas.

Radio Planning Tool
To carry out the radioelectric coverage analysis, the application RADIOGIS [6], which is a tool developed by the authors for the management and calculation of the radioelectric coverage of radio communication systems such as GSM, UMTS, LTE, TETRA, TDT, WiFi, etc., was used.RADIOGIS works on Windows PCs and is integrated with the GIS software ArcGIS 9.1 from ESRI.RADIOGIS has, among others, the following functionalities: a) Calculations of power, electric field, or power density radioelectric coverage while being able to select the propagation model to be applied: ITUR-526 [8] and ITUR-1546 [9] for rural environments; Okumura-Hata [10], COST-231 [11] and Walfisch-Bertoni [12] for urban areas.

Radio Planning Tool
To carry out the radioelectric coverage analysis, the application RADIOGIS [6], which is a tool developed by the authors for the management and calculation of the radioelectric coverage of radio communication systems such as GSM, UMTS, LTE, TETRA, TDT, WiFi, etc., was used.RADIOGIS works on Windows PCs and is integrated with the GIS software ArcGIS 9.1 from ESRI.RADIOGIS has, among others, the following functionalities: a) Calculations of power, electric field, or power density radioelectric coverage while being able to select the propagation model to be applied: ITUR-526 [8] and ITUR-1546 [9] for rural environments; Okumura-Hata [10], COST-231 [11] and Walfisch-Bertoni [12] for urban areas.b) Percentage threshold calculations using a vector layer containing municipal areas, roads, etc. c) Management of databases of sites, power density radioelectric coverage, power systems, measurement campaigns, etc.

Frequency Bands
For the existing TETRA network, the frequency band for RADIECARM is 380-385 MHz (uplink, UL) and 390-395 (downlink, DL), which is the band reserved for PPDR systems in Europe [13].
For the new LTE network, several possibilities are available for the frequency band.ECC Report 218 [2] develops the necessary conditions to create a harmonized European framework for the implementation of future BB-PPDR systems.This report proposes the concept of 'flexible harmonization' to enable an efficient implementation of BB-PPDR systems.The frequency bands identified as candidates for harmonization are: In our analysis, the frequency bands 699-716 MHz (UL) and 729-746 MHz (DL) were used.These frequency bands meet spectrum regulations for public safety users and broadband services laid out in Spain's National Table of Frequency Allocations.Moreover, they are E-UTRA operating bands according to the ETSI [14,15].

Digital Information
The Murcia Region (South-West of Spain) is 11,296 km 2 .The territory is organized into 45 municipalities, and each municipality delimits an area (municipal area) (see Figure 3).The geographical information has been obtained from the National Geographical Institute of Spain.A digital terrain model (DTM)-in a raster format [7]-with a 100 × 100 m cell size (see Figure 3)-was used for the radioelectric coverage calculations in rural environments, which represents a compromise between accuracy and computation time.We can also observe the municipal areas as a vector layer with an associated attribute table that contains information (name, extension, population, etc.) in Figure 3 [7].

Quality of Service (QoS)
The signal received in mobile communication presents wide random variations that can be modeled by introducing a statistical correction.This statistical correction defines a fade margin (FM).If the FM is equal to zero, radioelectric coverage is only guaranteed in 50% of the locations in the cell (defined in the DTM).The micro scalar quality is defined by the percentage of locations in the cell in which radioelectric coverage is guaranteed.This quality sets the fade margin that is considered in the link budget.
Assuming a log-normal distribution, the FM can be estimated by: where σ L (dB) is the variability of locations in the cell and K(L) is the normalized distribution abscissas for the percentage of locations (L).The K parameter is related with the inverse function of Gauss G −1 (L) by the expression: Appl.Sci.2019, 9, 250 For calculations, a typical value for the UHF band in rural environments is a shadowing standard deviation with locations at 6 dB.If we fix a micro scalar quality of 90%, the fade margin is: The macro scalar quality is the percentage of cells with radioelectric coverage within an area.This area could be delimited by a polygon, that represents a municipal or a regional area in our case.In our calculations, we assumed a macro scalar quality of 90% for the regional term and 85% for the municipal term.

Propagation Model
Many propagation models, such as the Longley-Rice [16], Bullington [17], Vogler [18], Luebbers [19], and Deygout [20] models, have been used for UHF band radio planning in rural environments in addition to the recommendations of the Radiocommunication Sector of the International Telecommunication Union (ITU-R) [8,9].All these models estimate the mean value of the received signal in each cell of the DTM, taking into account the terrain profile between the transmitter and the receiver.
In the radio planning tool used, propagation loss is evaluated for each terrain profile by: where L O is the free space loss, that can be calculated by: L Terrain irregularities is the diffraction loss caused by obstruction of the terrain.For rural environments, this propagation loss can be estimated using the recommendation ITU-R P.526 [8].L Close environment is the propagation loss taking into account the multipath phenomenon.This loss is directly related with the morphography in the vicinity of the mobile.In the GIS, a raster was created, with a resolution of 100 × 100 m, in which each cell has a value (in dB) representing the loss due to the type of environment (rural pine forests, suburban environment and urban environment).The Hata model [10] was used to estimate the value for suburban and urban environments.A measurement campaign, for the TETRA system (400 MHz), consisting of five routes, was carried out by the authors within a prototype-which included three base stations-to estimate the path loss for the rural (pine forest) environment.In Table 1, the obtained values for the TETRA system are shown.Moreover, in this table, the values for the LTE system (700 MHz), in which the losses for the rural environment were estimated by considering ITU-R P.833 [21], are also presented.These values were applied to the simulations performed in this study.Figure 4 shows a comparison between the propagation model and the measurements in one of the routes.Table 2 gives the mean error and standard deviation of the difference between the propagation model and the measurements for each route.

Link Budget
Table 3 gives the link budget for the TETRA and LTE systems and for the downlink (DL) and the uplink (UL).The link budget was calculated for a mobile receiver, in which the antennas are mounted on a vehicle, with a gain of 2 dBi for the TETRA system and 3.5 dBi for the LTE system.Also, in the base station, the antennas have an omnidirectional pattern with a gain of 7 dBi for the TETRA system and a directional pattern with a gain of 15 dBi for the LTE system.Radioelectric coverage was calculated for a rural environment, so a noise-limited scenario was assumed.
For the LTE system, the sensitivity is calculated by: where F is the noise figure (7 dB for the DL and 2 dB for the UL), Nrb is the number of resource blocks (25 for a bandwidth of 5 MHz for the radio channel), RB is the bandwidth of each resource block (180 kHz), and SNIR is the average signal-to-interference-and-noise ratio (5 dB) In the proposed UL for LTE, a real throughput of 4.5 Mbps can be achieved with a sensitivity of −100 dBm using a 16 QAM modulation, allowing the transmission of full HD (1920 × 1080) video streaming.The maximum path loss (L max ) is in agreement with Dunlop et al. [22] for the TETRA system and with Elnashar et al. [23] for the LTE system.As seen in Table 3, for both systems, the worst case is the UL, and the maximum path losses that can be compensated for are 144.8dB for the TETRA system and 136.3 dB for the LTE system.These values were used in the calculations for both systems.

Radioelectric Coverage Calculations
Using the radio planning tool, individual radioelectric coverage can be calculated for a site considering the maximum propagation loss allowed by the link budget.Then global radioelectric coverage is calculated taking into account the best server principle (the value of each cell covered by several sites is the best value).Each radioelectric coverage is stored in the GIS using a raster and a vector point layer.The raster had the same resolution as the DTM and the value of each cell was now the received power.The vector point layer also has an associated attribute table containing all the information used in the calculations: transmitter power, transmission loss, antenna gains, frequency, propagation model, etc.).This allows us to use all the facilities of GIS for managing and doing calculations with spatial data.
The percentage of radioelectric coverage can be easily calculated in GIS because the radioelectric coverage has been stored as a raster.The functionality of GIS [7] that allows operation between a raster and vector layer can be used to evaluate the macro scalar quality of service (defined in Section 2.5, using a raster of the global radioelectric coverage and the vector polygon layer of the municipal areas.The map of the macro scalar quality was also stored as a raster and can be represented and managed in the GIS environment.

Methodology for Planning the New LTE Network
The proposed flow-chart to plan the new LTE network can be observed in Figure 5.
Appl.Sci.2019, 9, x FOR PEER REVIEW 7 of 12 vector point layer.The raster had the same resolution as the DTM and the value of each cell was now the received power.The vector point layer also has an associated attribute table containing all the information used in the calculations: transmitter power, transmission loss, antenna gains, frequency, propagation model, etc.).This allows us to use all the facilities of GIS for managing and doing calculations with spatial data.
The percentage of radioelectric coverage can be easily calculated in GIS because the radioelectric coverage has been stored as a raster.The functionality of GIS [7] that allows operation between a raster and vector layer can be used to evaluate the macro scalar quality of service (defined in Section 2.5, using a raster of the global radioelectric coverage and the vector polygon layer of the municipal areas.The map of the macro scalar quality was also stored as a raster and can be represented and managed in the GIS environment.

Methodology for Planning the New LTE Network
The proposed flow-chart to plan the new LTE network can be observed in Figure 5. First of all, the LTE radioelectric coverage for each municipal area is calculated using the existing TETRA sites and the existing sites with other radiocommunication systems (not TETRA).In our case, this leads to 34 sites in total (16 TETRA base stations and the 18 existing sites) (Figure 3).

−
If the QoS (see Section 2.5) is fulfilled, the following optimization algorithm can be applied: Initially, individual radioelectric coverages were calculated for the available sites and are ordered in a table from the least to the greatest radioelectric coverage.Then, an iterative process for each site was conducted starting with the first element (the one with the least radioelectric First of all, the LTE radioelectric coverage for each municipal area is calculated using the existing TETRA sites and the existing sites with other radiocommunication systems (not TETRA).In our case, this leads to 34 sites in total (16 TETRA base stations and the 18 existing sites) (Figure 3).

-
If the QoS (see Section 2.5) is fulfilled, the following optimization algorithm can be applied: Initially, individual radioelectric coverages were calculated for the available sites and are ordered in a table from the least to the greatest radioelectric coverage.Then, an iterative process for each site was conducted starting with the first element (the one with the least radioelectric coverage) and the same sequential order (from top to bottom).In each iteration, the following steps are carried out: 1.
The element of the corresponding table is discarded.

2.
The global radioelectric coverage is calculated with the sites not discarded or eliminated.

3.
If the macro scalar quality is accomplished for the region and the municipal areas, the site discarded in step 1 is eliminated and, if not, the site is kept in the table.
-If the QoS is not fulfilled, this fact means that more sites are needed.Therefore, the sites which do not exist yet but can be built with the necessary infrastructure (telecommunication tower, electrical line, stand, etc.)-the location of which is estimated according to the dark zones observed in the global radioelectric coverage map-should be progressively added until the QoS is accomplished.Then, the previously mentioned optimization algorithm can be applied.

The Existing TETRA Network
The radioelectric coverage is calculated using the 16 TETRA base stations (Figure 6a).Table 4 gives the percentage of radioelectric coverage for each municipal area.As can be observed, the macro scalar quality is above 90% for the region and 85% for the municipal areas.
Appl.Sci.2019, 9, x FOR PEER REVIEW 8 of 12 3. If the macro scalar quality is accomplished for the region and the municipal areas, the site discarded in step 1 is eliminated and, if not, the site is kept in the table.

−
If the QoS is not fulfilled, this fact means that more sites are needed.Therefore, the sites which do not exist yet but can be built with the necessary infrastructure (telecommunication tower, electrical line, stand, etc.)-the location of which is estimated according to the dark zones observed in the global radioelectric coverage map-should be progressively added until the QoS is accomplished.Then, the previously mentioned optimization algorithm can be applied.

The Existing TETRA Network
The radioelectric coverage is calculated using the 16 TETRA base stations (Figure 6a).Table 4 gives the percentage of radioelectric coverage for each municipal area.As can be observed, the macro scalar quality is above 90% for the region and 85% for the municipal areas.

The New LTE Network
The methodology proposed in Section 2.9 was applied in our case study.First, the radioelectric coverage for the new LTE network which will take into account the mentioned 16 TETRA sites, as well as the 18 existing sites (34 in total), was calculated (see Figure 6b).
It should be noted that the number of 'dark zones' (blue color in Figure 6) is greater for the LTE network than for the TETRA network.Table 4 shows the percentage of radioelectric coverage for each municipal area.The macro scalar quality of 85% for the municipal area is exceeded in the majority of cases, except for six municipal areas (Caravaca, Jumilla, Lorca; Moratalla, Totana and Yecla) out of 45.Furthermore, the percentage of LTE radioelectric coverage for the region is 85% (below the actual macro scalar quality of 90% for the TETRA network).
At this point, according to the flow-chart of Figure 5, new sites must be added to improve the radioelectric coverage.With the aid of the GIS, twelve more sites, represented as blue triangles in Figure 3, were found in the detected 'dark zones'.These new sites would have to be constructed with the necessary infrastructure (telecommunication tower, electrical line, stand, etc.).They were chosen because they are accessible by path, and it is easy for them to receive electric power.The new radioelectric coverage was calculated with the total 46 sites (Table 4).Now, the macro scalar quality of service is accomplished for the municipal areas and for the region, 85% and 90%, respectively.
The next step was to apply an optimization algorithm to see if the same quality of service could be maintained with fewer sites.
As can be observed in Table 4, 39 sites is enough to accomplish the QoS.These results indicate that the number of sites is a factor of 2.4 higher for the LTE network than for the TETRA network if we want to ensure the same macro scalar quality for both systems.Figure 7 shows the LTE radioelectric coverage with 39 sites, in which the dark zones have been reduced with respect to the case with 34 sites (see Figure 6b).The next step was to apply an optimization algorithm to see if the same quality of service could be maintained with fewer sites.
As can be observed in Table 4, 39 sites is enough to accomplish the QoS.These results indicate that the number of sites is a factor of 2.4 higher for the LTE network than for the TETRA network if we want to ensure the same macro scalar quality for both systems.Figure 7 shows the LTE radioelectric coverage with 39 sites, in which the dark zones have been reduced with respect to the case with 34 sites (see Figure 6b).We can also analyze other situations of interest for the PPDR agencies, such as the radioelectric coverage in natural parks (represented as polygons in Figure 7).Table 5 shows the percentage of radioelectric coverage for each natural park.The percentage of LTE radioelectric coverage is less than 85% for 6 of 20 natural parks (assuming either 46 or 39 sites), although the percentage of radioelectric coverage for municipal areas is equal or above 85% (see Table 4).Therefore, we need, in this case, to add several sites to improve the radioelectric coverage in these six natural parks.We can also analyze other situations of interest for the PPDR agencies, such as the radioelectric coverage in natural parks (represented as polygons in Figure 7).Table 5 shows the percentage of radioelectric coverage for each natural park.The percentage of LTE radioelectric coverage is less than 85% for 6 of 20 natural parks (assuming either 46 or 39 sites), although the percentage of radioelectric coverage for municipal areas is equal or above 85% (see Table 4).Therefore, we need, in this case, to add several sites to improve the radioelectric coverage in these six natural parks.

Conclusions
The conversion of an existing TETRA network operating in the 400 MHz band to a new LTE network operating in the 700 MHz band has been analyzed in a real scenario by means of both simulations and measurements.The latter were used to tune the propagation models employed and, in this sense, the values of the obtained path loss for the different type of environments considered (morphographic correction) are offered for both TETRA and LTE systems.Moreover, the study of special scenarios such as natural parks has also been considered in the analysis.
The results show that, according to the real scenario considered, the number of sites needed in the LTE network-for a specific quality of service (90% for the whole region and 85% for municipal areas), is a factor of 2.4 higher than for the TETRA network.

Figure 1 .
Figure 1.(a) Telecommunication tower with antennas and stand and (b) TETRA base station equipment.

Figure 1 .
Figure 1.(a) Telecommunication tower with antennas and stand and (b) TETRA base station equipment.

Figure 2 .
Figure 2. (a) Four-wheel drive vehicle with the TETRA antenna and (b) TETRA Mobile station equipment.

Figure 2 .Figure 2 .
Figure 2. (a) Four-wheel drive vehicle with the TETRA antenna and (b) TETRA Mobile station equipment.

Figure 4 .
Figure 4. Comparison between measurements and the propagation model.Figure 4. Comparison between measurements and the propagation model.

Figure 4 .
Figure 4. Comparison between measurements and the propagation model.Figure 4. Comparison between measurements and the propagation model.

Figure 5 .
Figure 5.The proposed flow-chart to plan the new LTE network.

Figure 5 .
Figure 5.The proposed flow-chart to plan the new LTE network.

Table 1 .
Obtained path loss for the different types of environments considered.

Table 2 .
Mean error and standard deviation of the difference between the propagation model and measurements for each route.

Table 3 .
Link budget for the TETRA and LTE systems.

Table 4 .
Percentage of radioelectric coverage in municipal areas.

Table 4 .
Percentage of radioelectric coverage in municipal areas.

Table 5 .
Percentage of radioelectric coverage in natural parks.

Table 5 .
Percentage of radioelectric coverage in natural parks.