This action might not be possible to undo. Are you sure you want to continue?
Objectives After this chapter the student will be able to: • discuss the relations between re-use distance, traffic capacity and speech quality. • understand the different steps in the cell planning process. • define C/I and C/A and state their limits in GSM. • assign frequencies to different cells and calculate the received capacity. • understand the procedure for estimating the cell coverage.
5.1 INTRODUCTION .......................................................................... 2 5.2 STEPS IN THE CELL PLANNING PROCESS ...................................... 2 5.3 DIFFERENT CELL TYPES .............................................................. 4 5.4 INTERFERENCE AND ITS EFFECT ON THE RE-USE DISTANCE ......... 6 Co-channel interference, C/I ........................................................ 6 Adjacent channel interference, C/A.............................................. 7 5.5 FREQUENCY RE–USE .................................................................. 8 C/I affecting re-use distance....................................................... 11 5.6 CAPACITY CALCULATION FOR CELLULAR NETWORK ................ 12 Definition of traffic..................................................................... 12 Models for traffic calculations ................................................... 13 Dimensioning a cellular network ............................................... 16 5.7 COVERAGE CALCULATION IN A CELLULAR NETWORK............... 17
The following describes the steps involved and the tasks to be performed at each step. The difficulty is to plan a network that allows future growth and expansion. 5.1 Introduction Achieving maximum capacity whilst maintaining an acceptable grade of service and good speech quality is the main issue for the cellular network planner. Michael Quek 2 . in a single flow of events.2 Steps in the cell planning process Cell planning means building a network able to provide service to the customers wherever they are. re-use patterns.5. A cell planner is most likely dealing with the content of several of these boxes at the same time. This work can be simplified and structured in certain steps. System Requirements Define Radio Planning Initial Cell Plan Surveys Individual Site Design System Growth On-going Testing Launch of Service Implementation Different steps in the cell planning process This process should not be considered just as it is depicted. channel assignment and cell size calculations. This process is by no means complete or undebatable. For instance. Planning an immature network with a limited number of subscribers is not the real problem. the radio planning and surveying actions are interlinked in an ongoing iterative process that should ultimately lead to the individual site design. Wise re-use of site locations in the future network structure will save money for the operator. Some of these steps are performed frequently whilst others are more rare. interference problems. In this chapter we will look at issues to consider when cell planning: cell types. Normally the output from one box is the input of another. each operator has its own flowchart of processes.
parameter setting. power and PCM links. diversity. channel loading plan. BSC. choice of combiners.System Requirements: • Coverage for different customers in different environments • Traffic behaviour of customers in different regions • Phases of build-out • Quality of Service. fading margins for indoor. GOS and speech quality • Expansion and future investments • Limited available bandwidth Define radio planning guidelines: • Coverage and interference: which prediction model to use. interference and time dispersion. Check space for antenna mounting. BCCH and LAI. isolation. • Site survey: Pinpoint exact location with GPS. Michael Quek 3 . Plan for allocating frequencies. Antenna gain. Investigate physical necessities such as space for equipment. The ideal planned locations have to be searched for any suitable building. antenna types. MSC and the access network. tower or vacant lot that could be leased for a reasonable cost. Investigate other systems antennae and interfering transmitters. • • • • Individual site design and parameter setting: Radio engineers need to select best site location from the options available from the site acquisition department. Dimensioning of BTS. equipment type Surveys: • Radio environment survey: Investigate path loss. system balance. outdoor and in-car • Traffic planning: choice of models and processes • Frequency planning strategy: choice of reuse pattern and allocation of frequencies • Testing and optimisation strategy Initial cell plan: • Idealised overview of site locations and BTS configurations • Predicted composite coverage and interference map • Cell configuration. roof clearance (first Fresnel zone empty). direction and tilt and ERP need to be decided.
Adjust output power. power.3 Different cell types A cellular network is created by means of placing equipment in strategic places to guarantee a certain perceived Quality of Service. Grade of Service. location updating and to detect missing neighbour relations. but this is not cost efficient. 5. handover failures. set parameters. physical implementation of antenna directions or tilts or any other measures to counteract detected problems. termination equipment for PCM link. and to study trends. radio channels quality. interference.Implementation: • Install: BTS. Which cell types to use must be weighed against cost and expected penetration. analyses and optimisation: • System diagnostics: collect statistics in OMC. traffic distribution. lightning equipment and antennas. airconditioning equipment. Also to investigate problems reported by customers and to validate changes undertaken such as frequency re-tuning or parameter settings . call success rate. • Analysis of the results above. On-going testing. Idealistic then would be to place a base station in every street corner. Launch of commercial service: • When the network is operational a commercial launch can be made. and • Optimisation of parameters. access links statistics. dropped calls. Drive testing to detect blank spots and interference and to confirm correct call set-up. earth bar. handover. • Commissioning tests of BTS. • Drive testing to localise weak signal strength. time dispersion or other radio problems. Coverage Capacity Penetration Cost Spectrum Quality The choice of cell types affecting several posts on the scale Important when designing a network is to find a balance regarding which combination of these three types of cells to use: Michael Quek 4 . timers. MSC or BSC to analyse traffic behaviour.
Cell coverage shown as a hexagon The use of different types of cells on the same area introduces the concept of a hierarchical structure. guarantying good coverage.1 to 1 km. The base station is transmitting at low output power and the antennas could be mounted on walls or in the ceiling.g. The base station antenna placement is below the rooftops of the surrounding buildings.). A micro cell can maintain indoor coverage in the lower levels of a building. and these will form a hexagonal cellular pattern. where the major part of the radio waves is propagated along the streets. base station coverage areas or cells are shown as hexagons. boundaries between the base station cells will theoretically form straight lines. The main rays are propagated over the rooftops. Micro cells have a typical coverage range from 0. leading to increasingly complex handover relations and planning. Normally the site location is on a hilltop or a rooftop. Michael Quek 5 .a backyard e. Thus. Pico cells are used when the capacity need is extremely high in certain hot spots.Macro cells have a typical coverage range from 1 to 35 km. Pico cells supplies coverage in indoor environment (or possibly outdoors in environments physically distinctly limited . This is so because the system is designed to let the mobile always operate on the nearest or best base station. perpendicular to the connection lines between the sites. In the theoretical part of cell planning.
f1 C/I > 9dB Interferer. thus degrading system performance. Michael Quek 6 . both using the same carrier frequency. But when the value 9 dB was decided.4 Interference and its effect on the re-use distance The limited frequency band available introduces interference in the system. C/I ratios will be poorer. Carrier. frequency hopping is not used and in that case the value 12 dB should be used instead of 9 dB. Co-channel interference is the relation between the desired signal C and the undesired re-used signal I. Co-channel interference. f1 C I Co-channel interference Suitable values for this ratio are settled by evaluation by a large group of listeners as to what is acceptable speech quality. The values given in the GSM recommendation are: C/I ≥ 9 dB The margin for Rayleigh fading is already included in this value. This means that in areas where this cannot be guaranteed. In phase 1A in the GSM specification. Distance between two co-channel sites decrease. resulting in an increased overall interference in the network. frequency hopping was assumed.5. meanwhile the rush of new subscribers increases. The use of space diversity reception will improve the C/I properties of the system as the problem with fading dips is reduced. The following thresholds are defined in the specifications for a C (carrier) 20dB above the sensibility level. C/I Reusing an identical carrier frequency in different cells is limited by cochannel interference or C/I. as the cell planners are forced to build up a network by means of clusters and frequency re-use.
9 dB This means that we can allow the adjacent carrier frequency A1 to be up to 9 dB stronger than our desired signal C. are not ideal. The index in A1 denotes that we are referring to the first adjacent channel. C/A As the filters. adjacent frequencies will be used in neighbouring cells. C/A2 is not really an issue. the carriers will somewhat affect each other.It should be mentioned that co–channel interference requires not only channels to be assigned to different cells but also that they are actually in use simultaneously. In other words. The relation between the desired signal C from the correct carrier and the undesired signal A from the carrier 200 kHz away is called adjacent channel interference or C/A. The limit for this ratio in GSM is: C/A1 ≥ . Adjacent channel interference. This means that some of the energy of the adjacent frequency will leak into serving cell and cause interference. The same reasoning can be applied to the second adjacent channel. the one 200 kHz away. A C/A ≥ . This means that interference will be more of a problem during busy hours than at other times. the carrier 400 kHz away. In this case the second adjacent carrier frequency A2 can be up to 41 dB stronger than our desired signal C.9 dB C fx Adjacent channel interference C/A fx+200 kHz f When a frequency re-use pattern such as the 3/9 pattern is used. limiting each carrier to its domain of 200kHz. Michael Quek 7 .
This is true if we look at how efficiently the frequency spectrum is used in one spot. Code Time Freq Code Time Freq Frequency efficiency in a TDMA system versus a FDMA system A frequency used in one cell can be re-used in another cell at a certain distance. etc) is not necessarily more frequency efficient than the first generation of cellular systems (NMT.5.e. A1 A1 Frequency re-use Michael Quek 8 . digital AMPS.). the re-use distance can be shorter. NMT and AMPS both need roughly the same bandwidth for the same number of channels. This distance is called re-use distance. and the capacity increased. GSM needs 200 kHz to fit 8 physical channels. i. TACS etc. The advantage of digital systems is that they can re-use frequencies more efficiently than the analogue ones. AMPS.5 Frequency re–use The second generation of cellular systems (GSM.
A cellular system is based on re-use of frequencies. All the available frequencies are divided into different frequency groups so that a certain frequency always belongs to a certain frequency group. G F E A B C D Frequency groups forming a cluster The frequencies can be divided into different frequency groups. The frequency groups together form a cluster. but not re-used. This introduces the terms re-use patterns and re-use grids. A cluster is an area in which all frequency groups are used once. The most common re-use patterns in GSM is "4/12" and "3/9". A group of neighbouring cells using all the frequencies available in the system frequency band is called a cluster of cells. By repeating the cluster over and over again a cellular network can be built. D3 A1 A2 B3 B1 D1 A1 D2 C3 C1 A3 D3 A2 B3 B1 B2 C1 A3 C2 B1 D1 B2 C1 D2 A2 D2 C3 C2 D3 B3 B2 A1 C3 C2 D1 D2 C3 C1 A3 C2 A2 4/12 frequency re-use pattern Michael Quek 9 .
Frequency groups Channels A1 1 10 19 B1 2 11 20 C1 3 12 21 A2 4 13 22 B2 5 14 23 C2 6 15 24 A3 7 16 B3 8 17 C3 9 18 The channel numbers used within a cell are always 9 channels apart.4/12 means that the available frequencies are divided into 12 frequency groups. arising from adjacent frequencies in neighbouring cells. Michael Quek 10 . C1. B1. A2.. D1.. Example: Channel assignment of 24 frequencies in a 3/9 cell plan. A1 C1 B1 A1 A3 A2 C3 B3 C1 C3 A1 A3 B1 A2 C3 A3 C2 B3 B1 A2 C3 A1 A3 A2 C1 C2 B3 B2 B2 C1 C2 C2 3/9 frequency re-use pattern Other ways to build up a network exists besides the 4/12 and 3/9 reuse pattern. . 3/9 means that the available frequencies are divided into 9 frequency groups located at 3 sites. This is beneficial since the transmitter combiner must have a specified channel separation between the carrier frequencies to combine. D3. It also eliminates the risk of possible adjacent channels interference within a cell which could otherwise take place.. which in turn are located at 4 base stations sites. This assumes that the base station has three cells connected to it. Problems with C/A might appear in certain parts of a cell. The frequency groups are often assigned a number or name such as A1.
The co– channel distances in each of the two cases. where D=7. Variations of 10 dB between an analogue system and a digital are not unusual. D3 A1 2R R A2 B1 D1 D2 C1 A3 A2 B1 3R D2 B3 B2 A1 C3 C2 D3 B3 B2 C1 A3 A2 B1 D1 D2 C1 C3 C2 D3 B3 B2 A1 C3 C2 D1 D2 C1 A3 A2 C3 C2 Relation between R and a hexagon If N is determined. The C/I requirement that is decided or specified therefore decides which re-use pattern to use. which can suffer interference from all directions. From geometrical relations the ratio D/R is found to be: ___ D/R = √3N where R is the cell radius. The first step in order to determine ”N" would be to calculate the ratio D/R between co–channel cells.2 R 4/12: D = 6 R (As a reference. or sites. most analogue systems used the 7/21 frequency re-use pattern. are: 3/9: D = 5.9R. The lower number of frequency groups "N".) A sector–cell normally suffers interference from a smaller number of interfering cells than an omni. D (co–channel distance) can be calculated. Ultimately the coverage range from the base will be limited by Michael Quek 11 .C/I affecting re-use distance It is easy to see that at a given cell size the re–use distance will depend only on the number of frequency groups ”N". the shorter the re-use distance. Capacity will increase but the interference from surrounding neighbouring co-channel sites will also increase. and the more compressed the network will be. 3/9 and 4/12. which is used in the frequency re–use pattern.
Actually. The larger D is. the call intensity γ and the average duration τ of each call are essential. One way of defining traffic is this: ASUB = γ ×τ 3600 [E] ASUB = traffic from one subscriber γ = number of calls per hour per subscriber τ = average call duration Typical values could be: γ: τ: ASUB = 1 120 seconds 1× 120 × 3600 = 33 mE Michael Quek 12 . the less is the interference. we say that a mature cellular system is interference limited rather than noise limited. but also the capacity of the control channels needed to set up the call. To be able to offer a certain traffic capacity with a certain quality. Definition of traffic When estimating the traffic offered by one subscriber.6 Capacity calculation for cellular network One starting point when designing a cellular network is the estimation of traffic. some kind of statistical model to figure out the number of required channels is needed. Keeping this interference below a certain level is done partly by controlling the channel re–use distance. The result is the same: energy disturbing our desired signal in the very same frequency domain. Traffic theory gives us a chance to calculate on traffic. This disturbances might effect the possible re-use distance in the same way as C/I 5. This phenomenon is referred to as time dispersion. Thus. Reflected channel interference or C/R is the relation between the signal strength of the desired signal C and the reflected signal R. The calculations should be based on the busy hour of the network. When dealing with GSM we need to handle not only traffic capacity.these rather than by the noise. there is no difference in interference caused by co-channel interference or by reflected channel interference. and from the results base our dimensioning of the network.
i. • If we have an estimate for the total traffic to be handled by the system. also called the first formula of Erlang. n channels Offered traffic GoS Statistical model Served traffic Traffic model Erlang B. is the model most commonly used for this purpose. The probability of a call to be blocked is called Grade of Service or GoS. we need a statistical model to find out the number of traffic channels needed for the task. Therefore networks are designed for different probabilities of blocking. Offered traffic is the traffic offered by the subscribers wanting to use the system.e. and for the GoS. Models for traffic calculations First some definitions of different types of traffic. we cannot justify a network like that. that is the offered traffic. Rejected traffic is the traffic that is blocked due to lack of system capacity. One Erlang corresponds to one line being occupied 100% during one hour. no calls are blocked until all devices are busy The number of callers >> the number of devices Michael Quek 13 . • • Carried or served traffic is the traffic that is successfully carried or served by the system. • • Full availability. a total traffic of 33 E would be generated. This is the amount of traffic that would take place if the number of devices were so large so that no calls were rejected. ATOT = 1000 × 1 × 120 3600 = 33 E Traffic is measured in the unit Erlang. In this formula the following is assumed. Most GSM-operators use a GoS between 1 and 5 %.If the number of subscribers is 1000. To be absolutely certain that no calls are blocked. each subscriber would need his or her own private line or traffic channel! Of course.
• • • No queues. a blocked call will not be repeated Call attempts are randomly distributed Exponential distribution of call duration The relation between the three components.e. A and GoS. n. i. are here shown for certain values in the Erlang B table on the next page: Michael Quek 14 .
585 30.657 31.316 44.609 34.861 40.851 51.314 17.04167 0.522 15.808 34.559 54.378 27.0% 0.608 16.0% 3.15259 0.305 35.010 19.844 29.007 29.113 29.537 43.Grade of Service (GOS) 1.581 16.265 20.238 13.7998 11.086 52.333 75.6147 7.9090 2.916 32.746 20.174 11.0454 1.677 30.3702 6.802 25.412 33.182 14.215 36.059 11.787 39.507 26.216 20.598 21.0% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 0.01010 0.850 13.690 64.22347 0.150 21.129 29.8415 6.0% 2.6500 10.481 49.462 38.0% 2.66667 2.0% 20.096 42.6662 5.295 16.388 32.651 14.4871 9.755 14.011 43.230 12.0210 6.0923 1.680 31.456 38.758 34.534 44.6271 4.28155 0.381 48.614 33.913 46.432 34.249 16.677 16.036 13.461 13.7649 3.392 39.562 50.7378 4.108 39.437 11.640 32.5106 8.885 24.028 62.939 25.737 21.0% 4.5430 5.5246 2.773 26.434 34.932 22.062 23.301 32.800 1.715 38.5727 8.596 35.9354 3.473 12.526 31.557 39.81202 0.295 37.3608 1.5971 6.9299 2.1599 5.322 35.656 11.939 25.041 57.2830 5.7272 7.253 36.891 24.959 17.109 37.890 26.3994 1.0% 3.746 27.0840 5.125 16.2157 7.05263 0.237 31.995 27.381 41.189 17.483 19.2708 1.9667 8.500 17.529 26.656 32.189 2.579 17.543 35.776 27.165 45.711 28.505 18.498 31.342 71.997 14.4740 10.572 36.796 56.2977 10.7825 4.754 39.867 23.499 27.11111 0.565 38.222 14.255 41.25000 1.987 24.616 49.836 33.45549 0.284 20.7479 5.675 17.0000 1.844 26.896 15.2003 9.0573 2.333 13.933 42.353 19.848 24.132 18.252 28.784 22.505 11.337 21.0% The Erlang B table Michael Quek 15 .904 22.033 57.758 29.3692 8.718 54.6516 10.643 37.8284 10.59543 0.088 44.692 20.575 34.353 66.221 22.2497 3.679 69.643 32.534 37.952 12.660 79.624 33.7584 4.4798 5.270 58.698 29.152 32.435 27.885 15.166 30.997 31.2759 2.584 36.682 35.043 23.358 42.4015 8.0104 5.016 67.6571 1.1080 8.2302 7.985 20.909 23.533 45.544 12.468 32.0796 5.508 59.888 30.0% 2.379 56.8750 9.425 18.607 36.3280 7.336 43.004 37.03093 0.8760 6.191 22.5294 6.448 43.761 16.936 42.254 29.533 5.513 36.006 72.053 28.3517 8.718 30.503 35.725 24.721 28.827 23.985 41.9865 4.8954 6.158 22.3608 2.9501 8.524 38.0% 40.243 22.640 19.60221 0.720 28.608 33.0% 0.833 24.6072 7.0096 9.315 15.779 39.703 59.405 40.548 16.487 20.0% 5.838 13.8811 2.430 39.464 36.626 25.296 4.2185 2.056 52.036 14.419 13.0% 2.365 61.219 21.807 18.631 17.074 47.343 28.997 77.577 18.550 40.8349 9.031 12.7295 10.800 28.797 18.039 21.988 82.901 38.395 51.778 16.2589 1.324 80.174 30.357 34.383 19.755 44.368 12.216 42.081 30.018 40.941 30.840 35.943 21.857 12.02041 0.735 49.540 42.0764 7.099 23.651 19.392 20.385 14.414 46.652 40.500 14.771 31.3447 5.9603 3.545 41.484 13.5095 4.031 19.424 21.71513 0.644 10.5955 8.322 53.1086 6.892 3.0% 0.0000 3.5217 9.164 31.33333 0.1907 9.305 21.243 46.064 25.833 23.4612 5.323 40.5464 7.635 22.470 11.633 11.4300 9.990 24.192 27.322 47.772 24.541 26.954 19.844 27.613 18.470 15.121 37.5431 3.401 48.281 26.413 15.9452 4.147 47.680 44.140 23.86942 1.1276 3.975 41.0% 10.367 33.437 41.067 36.115 13.484 37.89940 1.647 29.048 22.045 14.062 40.543 33.5009 3.638 25.665 15.8035 9.155 38.491 12.38132 0.203 37.669 74.574 37.1410 7.914 25.6850 10.080 19.
0096 E can be handled in that case.30 33. the GoS required is 2 %.95 3. of channels 20 14 10 8 8 60 No. Michael Quek 16 . also called the second formula of Erlang.25 4. Another model is the Erlang C. the offered traffic we can take care of is found in the Erlang table. Those cells must together offer the system 33 Erlang. Dimensioning a cellular network The subscribers in a small town generates a total offered traffic of 33 Erlang.30 3. Instead. This division must be defined by the customer or made from reasonable assumptions. the total traffic is divided among the cells. The main difference is that in this so called queue model the subscriber will wait until a device is available independently of how long. This model will give us a slightly lower traffic capacity. Assume the task is to find the necessary number of channels per cell. which isn’t possible due to the correlation between cluster size and available frequency band.Example: If the number of available channels (TCH:s) is 15. of frequencies 3 2 2 2 2 11 Traffic distribution divided between different cells in an area As shown above.20 8. First. we assume that the coverage requirement indicates that 5 cells are needed. Let us assume the following traffic distribution: B C D E A Cell A B C D E Σ cells % Traffic 40 25 15 10 10 100 ATOT in Erlang 13. the Erlang table gives that 43 channels will be enough! That would require 6 frequencies in one cell.00 No. 9. in order to cover this traffic demand. traffic distribution over several cells results in more necessary channels than if all traffic had been concentrated to one cell. If the grade of service requirements during busy hour are 2%. with an acceptable grade of service of 2 %.
Michael Quek 17 .7 Coverage calculation in a cellular network Assuming that the traffic calculations has shown that the capacity is not a problem in the cell. 5.Unfortunately. propagation calculations are quite time consuming. Normally they would be handled by computer-based prediction tools. For this we need a model to be able to predict the propagation path loss Lp on the radio link between Base Transceiver Station (BTS) and Mobile Station (MS). from the most straightforward interference and traffic analysis. the number of TCHs needed will probably not correspond exactly to the number of frequencies allocated. to more complex simulations of the GSM system. In doing so. Urban areas Free space loss Knife edge diffraction Multipath propagation Shadowing Changes in the radio environment affect path loss The remaining available approach is therefore to use statistical distributions to describe the radio environment. after having calculated the traffic and found a channel requirement. Each new frequency carries 8 physical channels which all comes in a bunch at allocation. the MS experiences such a complex variety of superimposing and transitory shadowing. If done manually. During one single call. Then there is time to look closer at the possibility of having a coverage restricted system. most propagation models deal only with the prediction of the median value for Lp. its variability being accounted for with the use of different fading margins for different environments. absorption and scattering effects that there is no exact solution to the problem of finding Lp. Coverage predictions can be seen as the basic platform upon which the planning tool enables the Cell Planner to do all kinds of post processing analysis.
This is done by estimating how far out from the BTS that the radio waves can reach. trucks etc) as well as reflections from the surroundings. we have to start with balancing the system so that both uplink and downlink are equally strong. structures. we add margins to our calculations of the required signal strength at the cell border. since the MSs have a limited output power and would not be able to reach all the way back to the BTS. That is done by calculating an appropriate output power from the BTS. That gives us the possibility to “translate” that into a distance from the BTS to the cell border (which is a measure of the cell size we wanted from the beginning). and the required signal strength at the cell border is calculated. gives us mathematical relations between signal strength loss (i. At this stage it would be easy to just install a couple of BTSs with high output power. still strong enough for a MS on the cell border. so that we can find out how far out from the BTS we can have the cell border. to maintain calls and signalling procedures. by using a radio wave propagation model. To make sure that the MS will experience a good enough signal at all times.e coverage) everywhere. The problem would then be a weak uplink. Since the radio waves have to overcome all sorts of shadowing obstacles (buildings. we know the maximum allowed loss of signal strength from the BTS to the cell border. When the output power from the BTS is known. strong enough to give good signal strength (i. To be able to do that we need to specify how strong signals we would like the MS to receive on the cell border. All this is of course done with the help of a computer based prediction tool to be able to come up with an initial cell plan (third step in the cell planning process shown). we need to calculate the approximate size of the cells. Michael Quek 18 .To be able to find out how many cells that are needed to cover a certain area. the resulting radio signal will vary a lot. After that we would like to estimate the range of the radio waves. applicable for different cell types and environments.e “path loss”) and distance. with respect to the balance. Thus. considering the limited output power from the MS. These different models.