+ All Categories
Home > Documents > Educational cellular radio network planning software tool

Educational cellular radio network planning software tool

Date post: 22-Sep-2016
Category:
Upload: jmh
View: 215 times
Download: 1 times
Share this document with a friend
13
IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998 203 Educational Cellular Radio Network Planning Software Tool Fernando P´ erez-Font´ an, Member, IEEE, and Jos´ e Mar´ ıa Hernando R´ abanos Abstract— In this paper an educational software tool (CELLPLAN) is presented. This tool is used to illustrate to telecommunications engineering students the different steps typically followed in the planning of cellular radio networks. The software tool described in this paper simulates the procedures and algorithms implemented in professional planning tools in a simplified way. The major difference being that no actual terrain information (Digital Terrain Model) is used in the simulations but rather a “generic” terrain configuration is assumed. The availability of this simulation tool in the classroom gives fifth-year telecommunications engineering students the opportunity to tackle realistic cellular network design cases as if they were in a real engineering office. Index Terms—Cellular network planning, educational software tool, mobile communications. I. INTRODUCTION C ELLULAR networks [1] are nowadays one of the fastest growing telecommunications systems around the world. Networks following analog and digital standards are being deployed in almost every country. A well-known example is that of the GSM cellular system in Europe. There is a need for telecommunications engineering students to develop a solid background in this field. Telecommuni- cations engineering curricula take into account this fact and provide students with a range of subjects covering most aspects of this engineering field. In this paper, an educational software tool is presented that is being used in a graduate course on Mo- bile Communications at Vigo University Telecommunications Engineering School (Spain). Cellular planning is a rather complex issue in which a large number of concepts must be mastered in order to be able to optimize this type of networks in a highly competitive environment. An inadequate coverage, a high probability of blocked or dropped-out calls, low transmission quality, etc., will certainly make subscribers of a badly planned network turn to another service provider. Cellular planning involves a thorough knowledge of propa- gation effects in different types of scenarios: urban, suburban, wooded areas, open areas, etc. Propagation effects are ex- Manuscript received May 4, 1995; revised February 23, 1998. F. P´ erez-Font´ an is with ETS de Ingenieros de Telecomunicacion, De- partment of Communications Technologies, University of Vigo, Campus Universitario, E-36200 Vigo, Spain (e-mail: [email protected]). J. M. H. R´ abanosis is with ETS de Ingenieros de Telecomunicacion, Department SSR, Universidad Politecnica de Madrid, Ciudad Universitaria, E-28040 Madrid, Spain (e-mail: [email protected]). Publisher Item Identifier S 0018-9359(98)05722-7. tremely variable with the position along the mobile terminal route. This necessitates the use of statistical definitions for both coverage and interference quality. Teletraffic computations must also be performed in order to allocate to each base station a sufficient number of traffic radio channels to handle the demand in the area (cell) served by it. The traffic-handling capability of a base station is also expressed in statistical terms by defining an acceptable Grade of Service (GOS) or probability of blocked calls. Another aspect that requires attention is adequate selection of carrier frequencies used in every base station in order to keep interference below acceptable limits. As has been described earlier, cellular network planning is a complicated matter and, in some cases, students may lose track of the whole planning procedure if this is explained to them only using conventional classroom methodologies. That is the reason why the use of computerized tools was considered as complementary material in the teaching of cellular planning issues. Also, it is well known that cellular-phone operators use software planning tools to help in this complicated task. It was decided that an educational software tool was to be built implementing most procedures and algorithms found in such professional software tools. These tools are extremely expensive and make use of expensive Digital Terrain Models as input data. It was considered that such tools could not be afforded by the University and that simplified software packages would do just as well. The software tool described in this paper simulates the working procedures implemented in professional planning tools ([2]–[4], etc.). The availability of this software in the classroom gives the students the opportunity to study realistic cellular network design cases as if they were in a real engineer- ing office. The authors are not aware of a similar educational software tool following an approach like the one described in this paper. The first version of the program was fully developed under MATLAB 4 for WINDOWS. It was found that this envi- ronment was very useful for carrying out a feasibility study of such a software tool given the helpful menu generation, graphic, and statistical features built into MATLAB. Soon it was found that if large networks were to be simulated other alternatives should be sought. The software was optimized by writing some program routines in C language (.MEX files). With this approach, networks of up to 50 base stations could be handled within reasonable time spans: approximately, 15 to 25 min. 0018–9359/98$10.00 1998 IEEE
Transcript
Page 1: Educational cellular radio network planning software tool

IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998 203

Educational Cellular RadioNetwork Planning Software Tool

Fernando Perez-Fontan, Member, IEEE, and Jose Marıa Hernando Rabanos

Abstract— In this paper an educational software tool(CELLPLAN) is presented. This tool is used to illustrate totelecommunications engineering students the different stepstypically followed in the planning of cellular radio networks. Thesoftware tool described in this paper simulates the proceduresand algorithms implemented in professional planning tools ina simplified way. The major difference being that no actualterrain information (Digital Terrain Model) is used in thesimulations but rather a “generic” terrain configuration isassumed. The availability of this simulation tool in the classroomgives fifth-year telecommunications engineering students theopportunity to tackle realistic cellular network design cases asif they were in a real engineering office.

Index Terms—Cellular network planning, educational softwaretool, mobile communications.

I. INTRODUCTION

CELLULAR networks [1] are nowadays one of the fastestgrowing telecommunications systems around the world.

Networks following analog and digital standards are beingdeployed in almost every country. A well-known example isthat of the GSM cellular system in Europe.

There is a need for telecommunications engineering studentsto develop a solid background in this field. Telecommuni-cations engineering curricula take into account this fact andprovide students with a range of subjects covering most aspectsof this engineering field. In this paper, an educational softwaretool is presented that is being used in a graduate course on Mo-bile Communications at Vigo University TelecommunicationsEngineering School (Spain).

Cellular planning is a rather complex issue in which a largenumber of concepts must be mastered in order to be ableto optimize this type of networks in a highly competitiveenvironment. An inadequate coverage, a high probability ofblocked or dropped-out calls, low transmission quality, etc.,will certainly make subscribers of a badly planned networkturn to another service provider.

Cellular planning involves a thorough knowledge of propa-gation effects in different types of scenarios: urban, suburban,wooded areas, open areas, etc. Propagation effects are ex-

Manuscript received May 4, 1995; revised February 23, 1998.F. Perez-Fontan is with ETS de Ingenieros de Telecomunicacion, De-

partment of Communications Technologies, University of Vigo, CampusUniversitario, E-36200 Vigo, Spain (e-mail: [email protected]).

J. M. H. Rabanosis is with ETS de Ingenieros de Telecomunicacion,Department SSR, Universidad Politecnica de Madrid, Ciudad Universitaria,E-28040 Madrid, Spain (e-mail: [email protected]).

Publisher Item Identifier S 0018-9359(98)05722-7.

tremely variable with the position along the mobile terminalroute. This necessitates the use of statistical definitions for bothcoverage and interference quality. Teletraffic computationsmust also be performed in order to allocate to each base stationa sufficient number of traffic radio channels to handle thedemand in the area (cell) served by it. The traffic-handlingcapability of a base station is also expressed in statisticalterms by defining an acceptable Grade of Service (GOS)or probability of blocked calls. Another aspect that requiresattention is adequate selection of carrier frequencies usedin every base station in order to keep interference belowacceptable limits.

As has been described earlier, cellular network planning is acomplicated matter and, in some cases, students may lose trackof the whole planning procedure if this is explained to themonly using conventional classroom methodologies. That is thereason why the use of computerized tools was considered ascomplementary material in the teaching of cellular planningissues.

Also, it is well known that cellular-phone operators usesoftware planning tools to help in this complicated task. Itwas decided that an educational software tool was to bebuilt implementing most procedures and algorithms found insuch professional software tools. These tools are extremelyexpensive and make use of expensive Digital Terrain Modelsas input data. It was considered that such tools could notbe afforded by the University and that simplified softwarepackages would do just as well.

The software tool described in this paper simulates theworking procedures implemented inprofessional planningtools ([2]–[4], etc.). The availability of this software in theclassroom gives the students the opportunity to study realisticcellular network design cases as if they were in a real engineer-ing office. The authors are not aware of a similar educationalsoftware tool following an approach like the one described inthis paper.

The first version of the program was fully developed underMATLAB 4 for WINDOWS. It was found that this envi-ronment was very useful for carrying out a feasibility studyof such a software tool given the helpful menu generation,graphic, and statistical features built into MATLAB. Soon itwas found that if large networks were to be simulated otheralternatives should be sought. The software was optimized bywriting some program routines in C language (.MEX files).With this approach, networks of up to 50 base stations couldbe handled within reasonable time spans: approximately, 15to 25 min.

0018–9359/98$10.00 1998 IEEE

Page 2: Educational cellular radio network planning software tool

204 IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998

It must be pointed out here that professional planning toolstake much longer computation times to carry out the sametasks. In some cases, computation requires one full day, andcomputers are often left to run overnight. This is due to the factthat their propagation studies use detailed terrain informationin the form of Digital Terrain Models.

Presently, two versions of the original package are availableby contacting either author; one runs under MATLAB 4including some C routines as indicated above and the otherversion was written in TURBO PASCAL language under DOS.Both versions can be used in a very straightforward way sincethey provide pull-down menus where all commands to carryout the different phases of the planning process can be easilyfound (menus are available both in Spanish and English).Learning how to use CELLPLAN is a fairly easy task. Amenu-driven program like the one being described here greatlysimplifies the learning process which will take a couple ofhours in front of the computer after a general overview of theprogram features and its structure have been provided by theteaching personnel. Furthermore, already finalized examplesare available on the computer hard disk that can be loaded tosee what the final outcome of the program should look like.

Several alternatives are being considered at the momentfor the upgrade of the tool. One basic requirement is thatthe program should fully run under Windows. One possiblealternative is to write the whole package in C or PASCALlanguage. A preferred alternative is to use the compilationoptions of MATLAB 5. In this case, all the developmentadvantages offered by MATLAB 5 will still be available forthe easy introduction of new features to the program.

Additionally, new features available in second-generationcellular systems like GSM, for example, discontinuous trans-mission, power control, slow frequency hopping, etc., arebeing considered for implementation in new versions of thissoftware tool.

II. I NPUT DATA USED IN PROFESSIONAL

CELLULAR PLANNING TOOLS

When planning cellular networks such as AMPS, TACS,NMT, GSM, etc., propagation computations are of paramountimportance. For such calculations the basic input file requiredis one describing terrain irregularity. This type of file isknown asTerrain Database(TDB) or Digital Terrain Model(DTM). These files are grid-oriented, describing the terrainas a regular mesh of height samples. The resolution of thesedatabases may vary, with values ranging from 5050 to1000 1000 m .

Other input data files are those which describe the land-usage or environmental characteristics of the area being stud-ied: urban, suburban, rural, etc. This information is necessaryto evaluate the extra losses caused by the surrounding environ-ment. These additional losses are due to the fact that mobileantennas are low (1.5 to 3 m) compared to the surroundingenvironmental features, that cause blockage or shadowingevents when the mobile terminal travels behind a building, atree, or any other feature in the vicinity of the mobile terminal.

Environment information is contained in files knownas Land-Usage Databases(LUDB) or Morphostructure

TABLE ITYPICAL LAND-USAGE CLASSIFICATION

Fig. 1. Road and traffic demand maps.

Databases(MDB). In Table I, a typical land-usage classifica-tion [2] used in professional planning tools is shown.

Finally, maps containing the forecast traffic demand arenecessary in order to supply more radio channels to those areaswith higher demand (urban areas) than to areas with smallerdemand (rural areas). Files containing such information areknown asTraffic Databases (TrDB).

The educational software tool described in this paper doesnot consider a specific region where a cellular network is tobe deployed but, rather, a “generic” or “ imaginary” regionis assumed. In the examples presented throughout this papera study area of 120 120 km has been considered. Thisapproach does not require the use of TDB’s as input files. Thisgreatly reduces the data load to be handled by the simulationprogram. Still, the same principles and planning algorithmsmay be used without loss of generality.

In order to account for the terrain influence on radiopropagation, random laws are introduced which try to modelthe decay of the received signal with distance from each basestation.

The study area may be configured at user’s will by introduc-ing a road map file showing the main cities, roads, borders,administrative boundary lines, etc. Finally, by defining atraffic demand map or Traffic Database (TrDB) with differentdensities expressed in Erlangs per square kilometer, the studyregion is completely specified.

Fig. 1 represents an example of such a region where anetwork planning exercise will be carried out to illustrate thesoftware output plots (Section IV). In the figure, the main

Page 3: Educational cellular radio network planning software tool

PEREZ-FONTAN AND RABANOS: EDUCATIONAL CELLULAR RADIO NETWORK PLANNING SOFTWARE TOOL 205

Fig. 2. Program flow diagram.

Page 4: Educational cellular radio network planning software tool

206 IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998

TABLE IICELLULAR NETWORK PLANNING STEPS

roads, towns, and boundary (administrative) lines are showntogether with a traffic map.

In order to reduce the need to handle large data files, insteadof using a separate Morphostructure Database (MDB), theinformation contained in the Traffic Database is supplied to thepropagation simulation algorithms. In this way, for example,high-demand areas are associated to dense urban areas and lowtraffic demand areas are associated to open areas. Intermediatedensity areas are assumed to be suburban areas.

III. CELLULAR NETWORK PLANNING STEPS

In this section the main algorithms implemented in theeducational software are described. The program structure isalso presented in some detail. Cellular network design maybe divided into several phases which are listed in Table II. InFig. 2 a sketch of the flow diagram of the simulator is shown.

Several preliminary studies may be carried out in orderto define the adequate cell radius, transmitter power, etc., atthe start-up phase. Acellular spreadsheetis built into theprogram so that several radii, number of channels per cell,etc., are tested for a given average traffic demand.

At program start-up two menu options are available:

• Start the design of a NEW NETWORK.• Introduce a new base stations in an EXISTING NET-

WORK.

If the first option is selected, a road map and a traffic mapmust be input as well as a set of basic network parameters(Global Parameters, Table III). Then, the firstbase stationcan be defined. A set of base station specific parameters mustbe defined at this point (Table III). The process may continuewith the definition of another base station, the recomputationof all affected network parameters and the storage of thenew network configuration. Network planning can continuewith the deployment of new base stations some time laterin another planning session (LOAD EXISTING NETWORKmenu option). This simulates the temporal evolution of anetwork in which, as new subscribers use the network, newbase stations must be added.

Also, Mobile Station parameters have to be input to theprogram. Each network simulation can only be carried out fora single type of mobile terminal (vehicle-mounted, hand-heldportable, etc.).

A. Definition of a New base station

In the definition of a new base stationstep a location ischosen where no cellular coverage is available or a large trafficdemand is detected which cannot be handled by the currently

installed base stations, that is, the Grade of Service (GOS) isbelow the quality standard established for the network

GOS of blocked calls

Blocked call in this context means that all available trafficradio-channels in a given base station are each handling acommunication and, thus, no channel can be used to accepta new call.

Two methods may be used when selecting the sites for newbase stations. One follows as much as possible the geometricapproach of the classical cellular hexagonal geometry [1] withsome location tolerance. Following this approach, when greattraffic demand is found in a given area, thecell splittingtechnique can be used. The program overlays the classicalhexagon geometrical pattern on the simulated planning regionwhere the cellular network is being deployed. This overlaidinformation may be followed to position all new base stations.Several cellular patterns are available with cell clusters [1]with different numbers of cells, bothomnidirectional anddirective (sector) (Fig. 3).

The other possible approach would be to place base stationswithout following a regular pattern. This is in fact whathappens in most real-life situations. Planning engineers usuallytake into account the local characteristics of the area where thebase station will be installed using a good knowledge of thearea and their own engineering experience.

B. Propagation Studies

Once a new base station has been placed in the study areaand its parameters set (power, gain, antenna pattern, effectiveheight, etc.) apropagation study must be simulated. In orderto account for the terrain irregularity and other factors withoutusing aTerrain Database (TDB) file, some randomness mustbe introduced in the process so that cells with irregular shapesare generated, thus simulating different terrain configurations.

The propagation law used is the classical power law ap-proach [5]

where is the propagation loss in linear units andmay rangefrom for free space conditionsto nearly for urban areaconditions. The exponent is varied randomly inside the pro-gram (without user control) with azimuth in order to simulateterrain effects on propagation. Moreover, aGaussian variableis superposed on the path loss values obtained by using the

propagation law in order to simulate locations variabilitydue to shadowing/blockage effects and, thus, provide morerealistic propagation computations.

In order to account forclutter losses the Traffic Map isused as if it were aMorphostructure Database. Additionalmean losses associated to different traffic environments (ur-ban, suburban, rural, ) [5] are included in thecomputations.

The received power model for a given location follows theexpression

Page 5: Educational cellular radio network planning software tool

PEREZ-FONTAN AND RABANOS: EDUCATIONAL CELLULAR RADIO NETWORK PLANNING SOFTWARE TOOL 207

TABLE IIINETWORK PARAMETERS: GLOBAL, BASE, AND MOBILE STATION

Fig. 3. Overlaid cellular pattern for assistance in positioning new basestations.

where is a constant related to the transmitter power and thepropagation conditions, is the propagation exponent,represents the mean additional clutter losses, and isthe Gaussian variate mentioned above. The propagation law isschematically illustrated in Fig. 4.

For the computation of base station coverages and inter-ference levels,elementary surface elementsor predictionpixels are defined. In this case, and in order to lower thecomputational load, surface elements of 500500 m weredefined for the example shown in Section IV. Normally,greater resolutions [2]–[4] are used in practical planning tools(250 250, 200 200, 100 100, m ).

Received power levels are computed for each surface ele-ment as shown in Fig. 5. In the figure, the simulated terrainirregularity effects may be clearly observed. Fig. 6 showsanother cell shape produced by the randomization mechanismdescribed above.

For each base station areceived power fileis created whichspans over all prediction pixels in the study region (in theexamples, 120 120 m ). Fig. 7 illustrates the structure of thereceived power files for the different base stations introducedin the simulation together with the input files. Pixel-wiseoperations may be performed with this data arrangement notonly for propagation calculations but also for cell boundarycomputations, interference assessment, etc.

The received power computed for any base station at anysurface pixel may be interpreted as the median value of aGaussian distribution with alocations variability (standarddeviation) which depends on the type of environment

where the surface element is located. In the program describedin this paper, for simplicity, a single locations variability valueof 6 dB is assumed for all surface elements.

The consideration of the locations variability parameter al-lows the computation ofcoverageandcarrier-to-interference

values for different probability levels. For example, inFig. 5 a coverage plot is shown for 50% of locations and, inFig. 8, a coverage plot for 90% of locations is shown for thesame base station. It can clearly be observed how the coveragearea is dramatically reduced if a higher locations probabilityis specified.

C. Cell Boundaries

The following stage in the design of a cellular network isthe definition of cell boundaries. This is shown in Fig. 9(a)and (b) where it can be observed how the introduction of anew cell completely changes the shapes of other base stationspreviously installed in the area.

Cell boundary definition is a rather complicated issue. Itmay be defined in terms of the surface elements belonging toa given cell. This is the deterministic approach that is followedin the tool described here where each surface element belongsto only one cell.

However, each surface element (pixel) may be defined interms of the probabilities of a mobile in that particular surfaceelement being assigned to the different base stations [2] in thecellular network. In this way, a surface element could belongto several neighboring cells with different probabilities.

Other parameters influencing cell definition are, for exam-ple, thehand-off algorithm, thepower control mechanism,the type of terminal (vehicle-mounted, hand-held), etc. In theexamples presented in this paper a single mobile terminal type(vehicle-mounted) is used for all computed examples (SectionIV).

In this simulator a very simple approach to the identificationof cell elements (pixels) was adopted: a surface elementbelongs to the cell providing the highest received signal level(best server). This computation is carried out on a pixel-by-pixel basis (Fig. 7) by deciding, for every elementary surfaceelement (pixel), what base station provides the highest receivedpower.

D. Number of Radio Channels Required

At this point, thenumber of radio channels requiredmustbe assessed in order to guarantee theGrade Of Service(GOS)(blocking probability) established as the network quality stan-dard. The procedure implemented to carry out this computation

Page 6: Educational cellular radio network planning software tool

208 IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998

Fig. 4. Propagation model including random variations.

Fig. 5. Random cell shape produced by the program for 50% of locationscoverage probability.

Fig. 6. Another random cell shape produced by the program.

is the integration of the traffic densities in the TrDB for thenew cell. Once the total number of Erlangs for the new cell iscomputed, the number of channels required by the new cell iscomputed using the Erlang B formula [6] (Fig. 10).

Fig. 7. Input files, received power files, results files. Pixel-wise operations.

E. Frequency Assignments

The next step consists on thefrequency assignmentphase.Two options are again offered to the planning engineer, in thiscase, a student using the simulator. One option is to follow thestandard cellular approach [1] usingclusters, sets of channels,and channelgroups for large and small cells when splitting isused. An alternative is to useheuristic techniquesbased onthe evaluation of acompatibility matrix [2], [7]. The com-patibility matrix contains thedistanceexpressed in number ofchannels required for any cell pair. This matrix may be usedthen as an input to a heuristic frequency assignment algorithm[7]. Table IV illustrates the structure of a compatibility matrix.

Page 7: Educational cellular radio network planning software tool

PEREZ-FONTAN AND RABANOS: EDUCATIONAL CELLULAR RADIO NETWORK PLANNING SOFTWARE TOOL 209

Fig. 8. Random cell shape produced by the program for 90% of locationscoverage probability.

(a)

(b)

Fig. 9. (a) Cell boundaries with three base stations. (b) Cell boundaries withfour base stations.

represents the required separation in number of channelsfor cells and . represents the channel distance within thesame base station which will be dependent on the selectivity ofthe transmitter combining equipment. In this way, a separationof channels would mean that the same channels can be usedin both base stations, a separation ofwould mean that forthe two base stations considered a minimum separation of onechannel is required to keep interference below an acceptablethreshold, and so on.

Several criteria are available for the evaluation of thecompatibility matrix. If a protection ratio thresholdis defined, a thorough study would include a pixel-by-pixelevaluation of the ratio both in the Base-to-Mobile (down-link) and Mobile-to-Base (up-link) directions (Figs. 11 and

TABLE IVSTRUCTURE OF A COMPATIBILITY MATRIX

12). Simpler criteria may be used to evaluate the cell-to-cellcompatibility matrix elements in order to reduce computationtime. The program provides the following criteria:

• closest interfering point;• worst surface element;• average of the worst interference pixels;• pixel-by-pixel study (Figs. 11 and 12).

The heuristic frequency assignment algorithm [7] im-plemented in the software carries out a channel orderingprocedure in terms of the difficulty of a channel being assignedin previous iterations. Those channels having failed to beassigned are always assigned first when a new iteration starts.

F. Verification of Overall Network Quality

At this point, all planning steps have been completedexcept for averification of the network quality , i.e., thepercentage of locations for which adequate coverage andcarrier-to-interference levels are guaranteed and if the totalGOS is within acceptable limits. In the next paragraphs,the procedure for the evaluation of interference effects frommultiple sources is summarized in some detail.

Interference sources reaching an elementary surface elementare multiple (Fig. 13) and the evaluation of the statistics of thepower sum of all interference sources is not straightforward.

The parameter must be evaluated for all surface el-ements. A Gaussian distribution may be assumed for thereceived interference power from each interferer, where

is expressed in logarithmic units (dBm). The power sum,however, must be evaluated in linear units,(mW). Iffollows a Gaussian distribution, then will follow a log-normal distribution

(dBm) (mW)

What is sought is the evaluation of the statistics of

It is assumed that the overall interference expressed inlogarithmic units follows a Gaussian distribution

Normal

The detailed procedure for the evaluation of the statistics ofNormal can be found in [8].

Page 8: Educational cellular radio network planning software tool

210 IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998

Fig. 10. Pixel-by-pixel evaluation of the total traffic demand in a cell.

Fig. 11. Cell compatibility study. Up-link.

Finally, for the evaluation of the statistics of the carrier-to-interference ratio for each surface element, knowingthat the statistics of the wanted signal are also Gaussian

Normal

the following procedure is used:

Normal

where

and

This multiple interference algorithm is only evaluated for thedown-link direction for simplicity. Coverage and interferencemaps are shown after evaluation for all pixels in the studyregion. Also, overall network and cell-by-cell interferencestatistics are provided by the program. Other statistical studiesare also provided by the program; for example, the number oftimes the available channels are reused, etc.

IV. EXAMPLE OF CELLULAR NETWORK DESIGN

In this section several output plots produced by the programare presented. The input scenario was already depicted inFig. 1, where both the road map (plus towns, administrative

Page 9: Educational cellular radio network planning software tool

PEREZ-FONTAN AND RABANOS: EDUCATIONAL CELLULAR RADIO NETWORK PLANNING SOFTWARE TOOL 211

Fig. 12. Cell compatibility study. Down-link.

Fig. 13. Multiple interference statistics assessment on a pixel-by-pixel basis.

lines, etc.) and the traffic demand map are shown. In orderto place the base stations, a seven-cell cluster pattern withomnidirectional antennas was overlaid on the study area forguidance. For those sections of the planning area with largertraffic densities split cells were introduced (Fig. 3). Fig. 14shows the actual location of the input base stations and Fig. 15shows the best server plot for the different cell sites. Frequencyassignments were made using the heuristic technique describedin [7].

In Fig. 16, an interference plot map is presented overlaid onthe road map for reference. Computations have been made forthe whole network in the down-link direction. Fig. 17 presentsthe interference statistics for one cell in the form of a histogramwith the number of pixels in the cell with a givenvalue. Fig. 18 shows a whole network coverage plot overlaidon the road map for reference. In this example, coverage isguaranteed for more than 90% of the area considered and morethan 90% of the total traffic demand.

Fig. 14. Location of base stations.

The program may also produce channel usage histograms,tables with the number of channels and which ones are usedin each cell, as well as several other network performancestatistics.

V. BRIEF PROGRAM USAGE DESCRIPTION

In this section a brief description is given of the differ-ent menu options available in the program and how theyare related. This will further illustrate how the softwarepackage operates. Three captured computer screens showingexamples of the user-friendly menu system are illustratedin Figs. 19–21. The main menu options are:FILE, BASESTATION, NETWORK, and QUALITY. The most im-portant options available in the pull-down menu system aresummarized in Tables V-A and V-B.

All simulations must start and finish by selecting the mainmenu optionFILE . To introduce/remove/edit a base station

Page 10: Educational cellular radio network planning software tool

212 IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998

TABLE V-AMAIN PULL-DOWN MENU OPTIONS: FILE

Fig. 15. Cells in the simulated network.

the BASE STATION option must be activated. This optionincludes the realization of a propagation study from thenew base station. A modification in the cellular network, forexample, the introduction of a new base station, requires therecalculation of all major network parameters. This is achievedby selecting the option NETWORK which will preform thecomputation of cell boundaries, RF channel requirements,compatibility matrix, and frequency assignment.

Finally, to verify whether the coverage, interference andGOS quality is acceptable theQUALITY option must beactivated. This option allows the visualization of network mapsshowing coverage, interference and GOS levels for each study

area surface element (pixel) as well as tables summarizing theoverall and cell-by-cell quality of the network.

From this information, it may be concluded that parameterchanges in a given base station or the introduction of newbase stations may be required to achieve the desired networkquality. This is done by going back to theBASE STATIONoption in the main menu, thus going back to the beginning ofthe planning cycle:

BASE STATION NETWORK QUALITY

BASE STATION

A cellular network planning exercise will start with a blankstudy area where no base stations are yet defined. The studyarea may be configured by defining a number of roads, towns,and geographic and administrative limits. Finally, a trafficdemand map shall be input. This traffic demand map willinclude a background traffic level. Higher traffic demand areascan be defined around towns and roads by means of polygonsdrawn on the screen with the aid of the mouse.

Each network simulation will start by defining a number ofglobal parametersas indicated in the previous sections whichinclude the frequency band, the number of available channels,coverage, interference and GOS objectives, etc.

A simulation may be carried out in different sessions. In aninitial session, the global network parameters, maps, etc., and,possibly, some base stations may be introduced in the networkbeing planned. This planning exercise may be stored (CLOSENETWORK ) on the computer hard disk to be continued ata later time.

When a new base station (main menuBASE STATIONoption) is introduced in the network (its position, parameters,

Page 11: Educational cellular radio network planning software tool

PEREZ-FONTAN AND RABANOS: EDUCATIONAL CELLULAR RADIO NETWORK PLANNING SOFTWARE TOOL 213

TABLE V-BMAIN PULL-DOWN MENU OPTIONS: BASE STATION, NETWORK, AND QUALITY

Fig. 16. NetworkC=I > Rpth plot for two locations probability levels.

etc.) theNETWORK menu option has to be run again torecalculate all network base stations interrelations and, also theQUALITY menu option must be called up again to observewhat is the new network quality level in terms of coverage, in-terference, and GOS probability both at cell and network level.

The program implements a simpleModification Flag Sys-tem that keeps track of changes carried out in the network,

thus keeping track ofNetwork Configuration changes. Forexample, a base station power (EDIT BASE STATION )may be reduced to allow cell-splitting or a base stationmay be converted from omnidirectional to directive (sector)(EDIT BASE STATION ) or even a base station may becompletely removed (REMOVE BASE STATION ). Othermodifications that will trigger the Modification Flag System

Page 12: Educational cellular radio network planning software tool

214 IEEE TRANSACTIONS ON EDUCATION, VOL. 41, NO. 3, AUGUST 1998

Fig. 17. Cell C=I statistics.

Fig. 18. Coverage plot for two locations probability levels.

Fig. 19. CELLPLAN welcome window and main menu.

will be the introduction of a new traffic demand map (NEWNETWORK ).

These are all realistic events that may occur throughoutnetwork roll-out. For example, a traffic estimation made prior

Fig. 20. FILE, NEW NETWORK, LOAD EXISTING MAP menu option.

to network deployment will most certainly be different fromobserved values during network operation. Even during thelifetime of the network, traffic variations will surely happen.

Page 13: Educational cellular radio network planning software tool

PEREZ-FONTAN AND RABANOS: EDUCATIONAL CELLULAR RADIO NETWORK PLANNING SOFTWARE TOOL 215

Fig. 21. FILE, NEW NETWORK, LOAD EXISTING MAP menu option(continued).

When the modification flag is activated an indication ismade to the program operator that several items have tobe recalculated: cell boundaries, RF channel requirements,

(NETWORK ) and finally all network quality parameters(QUALITY ).

VI. CONCLUSIONS

An educational software tool for the simulation of theengineering steps in the design of a radio cellular network hasbeen presented. Future telecommunications engineers have theopportunity of simulating the complete process in the designof a cellular network since the tool provides the most commonfeatures found in professional planning tools used by networkoperators and consultants. The tool implements relevant al-gorithms which radio engineers must be familiar with such asmultiple interference statistics evaluation, compatibility matrixdevelopment, and heuristic frequency-assignment techniques,classical cellular layouts, propagation statistics, tele-traffictheory, etc. This tool is being successfully used in a graduatecourse on mobile communications at the TelecommunicationsEngineering School, University of Vigo, Spain.

REFERENCES

[1] V. H. Mac Donald, “The cellular concept,”Bell Syst. Tech. J., vol. 58,no. 1, pp. 15–41, Jan. 1979.

[2] M. Kruger and R. Beck, “GRAND—A program system for radionetwork planning,”PKI Tech. J., vol. 1, pp. 7–12, 1991.

[3] A. Bajwa, “Cellular radio planning tools,” inCellular Radio Systems, D.M. Balston and R. C. D. Macario, Eds. Norwood, MA: Artech House,1993, ch. 11.

[4] J. Kaarre and T. Kajamaa, “MONICA. A program for cellular networkplanning and data management,”Telecom Finland (Mobile TelephoneServices), Aug. 13, 1990.

[5] H. Hata, “Empirical formula for propagation loss in land mobile radioservices,”IEEE Trans. Veh. Technol., vol. VT-29, pp. 317–325, 1980.

[6] W. C. Y. Lee, Mobile Communications Design Fundamentals.NewYork: Wiley, 1993, ch. 8.

[7] F. Box, “A heuristic technique for assigning frequencies to mobile radionets,” IEEE Trans. Veh. Technol., vol. VT-27, pp. 57–64, May 1978.

[8] S. C. Schwartz and Y. S. Yeh, “On the distribution and moments ofpower sums with log-normal components,”Bell Syst. Tech.nical J., vol.61, no. 7, pp. 1441–1463, Sept. 1982.

Fernando Perez-Fontan (M’95) received the telecommunications engineer-ing degree from the Polytechnic University of Madrid, Madrid, Spain, in 1982and the Ph.D. degree from the same university in 1992.

He has been with the Department of Communications Technologies, Uni-versity of Vigo, Spain, since 1988. His main interests are in the field of radiopropagation modeling for terrestrial-mobile and land-mobile satellite systems.Currently, he participates in different European Space Agency Projects and inthe Euro-COST 255 Action “Propagation modeling for new SatCom servicesat Ku-band and above.”

Jose Marıa Hernando Rabanosreceived the telecommunications engineeringdegree from the Polytechnic University of Madrid, Madrid, Spain, in 1967 andthe Ph.D. degree from the same university in 1970.

He was with the ITT Research Laboratories (Madrid) from 1967 to 1969.From 1970 to 1977, he was with the Communications Department of IberiaAirlines of Spain, where he was engaged in the planning and design ofland-mobile and air-to-ground radiocommunication networks. In 1977, hereturned to the Polytechnic University of Madrid as a Full Professor in theRadiocommunications Department, where he has been working in the fieldof cellular network planning and the development of computerized planningtools for digital cellular networks.


Recommended