3 Radio planning with Atoll. The purpose of this paper is to provide the plan and design a UMTS mobile communications network to give coverage to the town of Seville. We used the software tool Atoll radio planning and simulation, developed by the company Forsk. With the help of this tool will determine the design parameters of the network and relevant simulations will be performed to verify that the objectives have been achieved quality.
3.1 About Atoll. Today is no longer regarded the implementation manual or any programming of all necessary calculations for radio planning as described in Chapter 2 of this document.In a professional environment they are always planning tools, except in very simplified. ATOLL is a radio planning environment based on windows, easy to use, supports wireless carriers throughout the lifetime of the network. From initial design to the optimization phase and during the various extensions [2]. More than an engineering tool, ATOLL is a technical information system open, scalable and flexible that it can be easily integrated into other telecommunications systems, increasing productivity and reducing development time. ATOLL allows a wide variety of deployment scenarios. From a single server, up configurations using parallel and distributed computing. The main features of Atoll are: · Advanced properties in network design: a tool for calculating propagation of high-performance, multi-network support and hierarchical traffic shaping, and automatic frequency planning and network optimization codes.It supports GSM / TDMA, GPRS, EDGE, IS-95 CDMA, W-CDMA / UMTS, CDMA 2000. Allows network planning technologies (GSM / UMTS, GSM / GPRS, CDMA/CDMA2000 ...). · Open and flexible architecture: it supports multi-user environments through architecture innovative databases that can share data, manage the integrity of the data and easy integration with other telecommunications systems.Allows the integration of proprietary modules (AFP propagation models) through a set of programming interfaces (APIs). It also allows the integration of macros. · Parallel and distributed computations: ATOLL allows the distribution of computation among multiple workstations and supports parallel
computations in multiprocessor servers, dramatically reducing the time of simulation and prediction, taking full advantage of hardware. · Art GIS, geographical data ATOLL supports multi-format and multiresolution and integration with GIS tools. Allows loading complex databases and display geographic information interactively with multiple layers, including engineering studies and prediction.Includes raster and vector editor. ATOLL is composed of a core module that can add modules such as UMTS module (allowing projects CDMA / CDMA 2000) specifically for the analysis and network planning W-CDMA/UMTS, the Measures module allows you to import and manage specific measures CW or test data mobile routes, Module Automatic Frequency Planning for the optimization of frequency plans GSM / TDMA and Microwave Planning module. This module allows users to plan and analyze microwave links. The advantages for our purposes is obtained from this application are based mainly on three aspects: · Allows us to have databases of high resolution topographic and access to them for terrain profiles and data to be used for calculations of propagation. · We can use methods of predicting the radio propagation more elaborate and much more laborious calculations, which would be impossible to perform manually. · It also allows us to have databases with existing or planned equipment.This makes it easier to compare different potential sites, antenna height, power equipment, etc. We have therefore a much higher range of possibilities and simplifies the process of network optimization. Atoll is based on digital terrain maps.The program can perform calculations on information extracted from these maps and databases that the engineer generates information on the network. Maps, databases and the results of these calculations are grouped into program files called "projects."
3.2 Traffic modeling. The first objective is to model in some way the traffic generated by the user population of the city of Seville [1] [2], [7]. We create a UMTS-type project (File | New) by selecting the template UMTS HSDPA. The first is to import the maps for the city of Sevilla (File | Import), select the index files of different folders that are grouped charts: Heights (map type altitudes) Clutter (clutter type classes) , Ortho (image) and Vector (lineare).
The resolution of the maps that we use is 25 m, which in principle is sufficient because the target area topography is fairly uniform and regular. The map is a map of heights and contains altimetry and topographic relief of the work area.The information contained in this map is used for the calculation of coverage and spread. Altimetry map we use for our study is shown in Figure 10.
Figure 10: Map of altimetry Seville. The clutter map is the map of land uses and in it, each type corresponds to a color field.The clutter that we will use is shown in Figure 11.
Figure 11: Map of land use (clutter classes) in Seville.
As shown in the legend, in the case of Sevilla have 12 types of zones: the open (OPEN), water (INLAND_WATER), residential (RESIDENTIAL), urban average (MEAN_URBAN) urban sprawl (DENSE_URBAN), buildings (BUILDINGS), village (VILLAGE), industrial (INDUSTRIAL), opened in town (OPEN_IN_URBAN), forest (FOREST), parks (PARKS) and dispersed urban (SCATTERED_URBAN). Ortho map is simply an aerial photo of the city. Is shown in Figure 12:
Figure 12: orthophoto map type.
Finally, the map identifies Vectors roads, rivers, railway lines, etc. Vectors map we will use is shown in Figure 13.
Figure 13: Map type Vectors of the city.
The layers of different maps overlap each other. Order can be changed by moving the mouse for almost all visible simultaneously. We will arrange to appreciate all the time clutter maps, orthophoto and vectors. The result of this overlap map shown in Figure 14:
Figure 14: Overlay of all city maps.
To model the traffic generated by the city are going to define user profiles, and each one will assign a number of UMTS services with certain parameters that indicate the user traffic generated by each service.We are only going to include in the model of services: voice, MMS, Internet access and video conferencing. It was not deemed necessary to modify the default values for these services Atoll, as are typical for UMTS planning in cities. The service features are included in Table 1. Service Name R99 Bearer Service Type Soft handoff allowed Priority Factor activity in the UL Factor activity in the DL Average date rate in the UL Average date rate on the DL Lost by the body
Voice
MMS
Internet
LCD12 Circuit mode
UDD64 Packet mode
Yes 2 0,4 0,4 12.2 kbps 12.2 kbps 3 dB
No 0 0,75 0,75 64 kbps 64 kbps 0 dB
UDD384 Packet mode No 0 0,75 0,75 64 kbps 384 kbps 0 dB
Table 1: Characteristics of UMTS services.
Video conference LCD64 Circuit Mode Yes 1 1 1 64 kbps 64 kbps 0 dB
These services can be obtained from different types of terminals. We will consider two different types of terminals: mobile phone and PDA. The terminal characteristics are those that have default Atoll, and are listed in Table 2. Terminal Minimum power Maximum Noise Type (dBm) power (dBm) Figure (dB) Telephone -50 21 8 PDA
-50
25.
Active set size 3
7
1
Table 2: Characteristics of UMTS terminals. User profiles with their services and associated terminal types listed in Tables 38.These values are set with reference to other studies dimensioning of UMTS networks to which access has been [1], [4]. ·
Adolescent (10-20 years):
Service
Terminal Calls per Type hour
Voice
Telephone
Mobile Telephone Mobile Access Telephone The Internet Mobile Video Telephone Conference Mobile MMS
Call duration Volume of data Data volume in in the UL DL (Kbytes) (sec) (Kbytes)
0,25
250
-
-
0
-
150
150
0
-
200
6.000
0,005
125
-
-
Table 3: Traffic generated by the user Adolescents.
· Service Voice MMS
Young (20-30 years). Terminal Calls per Call duration Volume of data Data volume in Type hour (Sec) in the UL DL (Kbytes) (Kbytes) Mobile 0,25 275 Phone Mobile 0 200 200 Phone
Internet Access Video Conference
Mobile Phone Mobile Phone
0
-
300
7.000
0,005
150
-
-
Table 4: Traffic generated by young users. ·
Middle-aged (30-50 years).
Service
Terminal Calls per Type hour
Voice
Mobile Phone MMS Mobile Phone Internet Mobile Access Phone Video Mobile Conference Phone
0,2
Call duration Volume of data Data volume in in the UL DL (Kbytes) (sec) (Kbytes) 200 -
0,005
-
100%
100%
0
-
200
6.000
0,025
100%
-
-
Table 5: Traffic generated by the user Median age. ·
Middle age (50-65 years).
Service
Terminal Type Mobile Phone Mobile Phone
` MMS Internet Access Video Conference
Mobile Phone Mobile Phone
Calls per hour 0
Call duration Volume data in Data volume in (sec) the UL (Kbytes) DL (Kbytes) 120 -
0,001
-
100%
100%
0,0025
-
200
6.000
0,00125
60
-
-
Table 6: Traffic generated by the user age.
· Service
Elderly (+65 years). Terminal Calls per Type hour
Call duration Volume of data Data volume in in the UL DL (Kbytes) (sec) (Kbytes)
Voice
Mobile Phone MMS Mobile Phone Internet Mobile Access Phone Video Mobile Conference Phone
0,05
60
-
-
0,0005
-
100%
100%
0,00125
-
100%
3.000
0,00005
30
-
-
Table 7: Traffic generated by the user person further.
· Service
Business Person. Terminal Calls per Type hour
Voice
Mobile Phone MMS Mobile Phone Internet Mobile Access Phone Video Mobile Conference Phone Voice PDA MMS PDA Internet PDA Access Video PDA Conference
0,5
Call duration Volume of Data volume in (sec) data in the UL DL (Kbytes) (Kbytes) 350 -
0
-
200
200
0,25
-
500
10.000
0
200
-
-
0,5 0 0,25
350 -
200 500
200 10.000
0
200
-
-
Table 8: Traffic generated by the user person business.
The next step for modeling the traffic generated by the city is to define a series of "environments" type, each of which will assign a population density of users associated with their mobility. Later on the map available generate an environment map, which is only noted on the map to that type of environment is for each pixel of the map. The types of mobility (Table 9) are those set by default Atoll, as they are considered typical values of UMTS in cities. Average speed
mobility rate (Km / h)
Eo / Io (dB)
Threshold HG-SCCH Ec / Nt (dB)
Pedestrian 50 Km / h 90 Km / h
3 50 90
-14 -14 -14
-9 -9 -9
Table 9: Types of mobility
And finally we define the environments. Each environment is characterized by a series of pairs "user profile" mobility "and a population density associated with each of them. Environments are defined as set out in Table 10. The densities were chosen by reference to demographic studies which have been accessed [7], [11]. Type of environment (ab/Km2) Open Urban Dense urban Residential Industrial Great Buildings
Population density (hab/Km2) Density of subscribers 400 20000 30000 5000 10000 40000
100 4000 6000 1000 2000 8000
Table 10: Types of environments from the city of Seville.
Is to size the network assuming that pays a 20% of the inhabitants of the city. Percentage is quite optimistic, which may take a long time even achieved or not achieved, but ensures that the network does not saturate easily. Then we estimated the density associated with each environment for each user group, again taking as reference demographic studies of the National Institute of Statistics [11].The results are shown in Table 11.
Type Teenage environment Open 8 Urban Urban dense Residential Manufacturing Buildings
Young
Medium
older
other
Business
21
39
21
9
2
eight hundred. 1200
1.200
700
75
1800
eight hundred. 1200
900
100%
200
275
200
150
25.
75
400
1000
400
75
50
1.050
1.600
2.400
1.600
1.200
150
425 eight hundred. 150
Table 11: Densities and types of users associated with Sevilla environments.
Finally, we must define what percentage of each user densities associated with the environment presented by each type of mobility. For this open environment User type Mobility Teen Young Median age older other Business
is shown in Table 12: Pedestrian 50 Km/h 2 3 7 7 13 13 7 7 3 3 0 1
90 km/h 3 7 13 7 3 1
Table 12: Types of users and mobility associated with the open environment
Table 13 shows what we have estimated for an urban environment: User type/Mobility Pedestrian 50Km/h 90km/h Teen 375 25 25 Young 700 50 50 Medium 1000 100 100 Older 700 50 50 Other 40 40 620 Business 50 13 12 Table 13: Types of users and mobility associated with the urban environment.
For a dense urban environment has been a percentage of subscribers in much lower vehicle, being mainly the old town area, which is intended to restrict vehicle access in the near future.Densities associated with the binomial type of user-mobility are shown in Table 14: User Type /Mobility Teen Young Median Older Other Business
Pedestrian 780 1170 1760 1170 880 95
50Km/h 10 15 20 15 10 3
90km/h 10 15 20 15 10 2
Table 14: Types of users and mobility associated with dense urban environment.
For the residential environment are also considered low densities for cases 50 km / h and 90 km / h, as they are considered low-traffic areas.The associated densities are given in Table 15: user type /Mobility Pedestrian 50 Km / h 90 km / h Teen 140 5 5 Young 180 10 10 Middle 250 13 12 Older 180 10 10 Other 140 5 5 Business 20 3 2 Table 15: Types of people associated with residential mobility.
All these parameters can be completed in the UMTS parameters folder in the data tab of the browser window.You can delete and add entries for folders: Environments, User Profiles, Terminals, Mobility Types, Services and within each entry you can change various settings for each input. The next step is to create a traffic map. To do this, on a digital map of Seville we will define a number of areas and each of them we assign one type of environment (environment map or raster). The map of environments we will generate a similarity of map of land uses which have the city of Seville.The land use map or clutter classes each zone shows a different color.
To create a traffic map Atoll Geo select the tab of the browser window, create a new road map, scenario-based or raster, and we mark on the map kind of environment that belongs to each zone. The result is shown in Figure 15
Open Residential Urban Dense urban High buildings Industrial estates Parks
Figure 15: Map of surroundings of the city of Seville.
3.3 Propagation model. It will use the propagation model Cost-Hata. Hata formula is specially designed for applications in mobile communications in any environment (COST231 is only for urban environments) and on the other hand, the Okumura-Hata method is only for frequencies below 1500 MHz Cost-Hata (or Hata, COST231) is a variation of the Hata formula for systems operating at 1,800 MHz and 2,000 MHz [4], as is the case at hand. Propagation Models folder in the Modules tab of the browser window assign a different formula for each type of clutter map area.
The allocation formula is that of Table 16: Zone Type Field (OPEN) Water (INLAND_WATER) Residential (RESIDENTIAL) Urban average (MEAN_URBAN) Urban sprawl (DENSE_URBAN) Buildings (BUILDINGS) Pueblo (VILLAGE) Industrial (INDUSTRIAL)
Cost-Hata formula Rural (open area) Rural (open area) medium-sized city and suburban Metropolitan Center Metropolitan Center Metropolitan Center medium-sized city and suburban Metropolitan Center
Open city (OPEN_IN_URBAN) Forest (FOREST) Parks (PARKS) Dispersed urban (SCATTERED_URBAN)
Rural (almost open) Rural (almost open) Rural (almost open) medium-sized city and suburban
Table 16: Allocation of Cost-Hata formulas to different types of environment.
The terms set out in the Atoll database for this method are: ·
Metropolitan Center:
Lu = 49.3 + 33.9 log f - 13.82 log Hb + (44.9 to 6.55 log Hb) gives log (M r) = (1.1 log f - 0.7) H r - (1.56 log f - 0.8) Total = Lu - a (H r) ·
Medium-sized city and suburban:
Lu logf = 46.3 + 33.9 - 13.82 logHb + (44.9 to 6.55 logHb) logd to (H r) = (1.1 logf - 0.7) H r - (1.56 logf - 0.8) Total = Lu - a (H r) ·
Rural (almost open):
Lu logf = 46.3 + 33.9 - 13.82 logHb + (44.9 to 6.55 logHb) logd to (H r) = (1.1 logf - 0.7) H r - (1.56 logf - 0.8) Total = Lu - a (H r) - 4.78 log 2 logf f + 18.33 - 35.94 ·
Rural (open area):
Lu logf = 46.3 + 33.9 - 13.82 logHb + (44.9 to 6.55 logHb) logd to (H r) = (1.1 logf - 0.7) H r - (1.56 logf - 0.8) Total = Lu - a (H r) - 4.78 log 2 logf f + 18.33 - 40.94 Finally, define Predictions folder as the default method of propagation Cost-Hata with a resolution according to the resolution of the maps (25 m) and a terminal height of 1.5 m. This value for the height of the terminal is a typical value used for such studies and that implies that all active users are at ground level, ie in the worst case (further away from the base station) .
3.4 Network equipment. We will introduce information about the technical characteristics of the computer in your network. These specifications pertain to the equipment described in Chapter 4. We will try to model with these teams Atoll as realistic as possible so that the results of the simulations are close to reality as possible.
3.4.1 Antennas. The description of the antennas are going to use is found in paragraph 4.1.4.1 of Chapter 4. Atoll contains a database with some antennas defined by default. We will create a new antenna from scratch, which is as close as possible to our actual antenna. To do this we create a new folder antenna Antennas Data tab of the browser window.The characteristics of the antenna set are shown in Table 17. The patterns of horizontal and vertical filing of the antenna are shown in Figures 16 and 17 respectively. Name Manufacturer Gain Power Tilt Beamwidth maximum frequency Minimum frequency
UD01P_D18BB Kathrein 18 dBi 4º 63 º 2,170 MHz 1920 MHz
Table 17: Properties of the antenna Atoll.
Figure 16: horizontal radiation pattern of the antenna UD01P_D18BB in Atoll. As described in Chapter 4, the antenna has a beamwidth of 63 ° in the horizontal plane (3 dB drop at 63 º) the attenuation is 10 dB at 120 º and the attenuation of the lateral lobes (90 º) is 20 dB (see Figure 16).
Figure 17: Radiation pattern of the antenna vertical UD01P_D18BB in Atoll. On the vertical beamwidth is 6.5 degrees and has introduced a power tilt 4 º (see Figure 17).
3.4.2 Base Station. The base station model chosen is the IN-60 from Nortel, whose main characteristics will be found in Chapter 4. The characteristics of the base station is included in Atoll in the corresponding deployment template. In the radio toolbar, select manage staff, make a copy of an old template and fill it with the specifications of our base station. The selected parameters are those of Table 18: Number of sectors Antenna model B 2 Frequency Band Height
3 UD01P_D18B ,170 MHz 30 m
base station Noise figure Pilot Channel Power SCH Power Power other CCH AS Threshold Maximum power Maximum load on the DL (peak) The maximum load on the UL Maximum date rate per user at DL Maximum date rate per user at UL Maximum number of CEs in the DL channel Maximum number of CEs in the UL
5 dB 33 dBm 21 dBm 30 dBm 5 dB 43 dBm 75% 50% 1,000 Kbps 1,000 Kbps 256 256
Table 18: Table of characteristics of the base station Atoll.
3.5 Deployment planning. Once we have modeled the traffic of the city of Seville can begin to locate the sites and have run simulations to achieve quality objectives. In principle we will look quality objectives in Table 19: Service Probability of service denial or delay Voice 2% MMS 5% Internet access 10% Video Conference 2% Table 19: Quality objectives. We set a target of availability of Voice and Video Conferencing as telephone networks are usually designed for a 2% chance of rejection. We have set a quality goal of 5% for MMS because it has a lower priority than those of the services operating in circuit mode (it is considered less critical) and not a delaysensitive service. Internet access service is the lowest priority and is also the most penalized other services, it is likely therefore to be the most likely to be rejected by the network and we may be difficult to obtain high levels of availability . We will begin the deployment of sites using the available templates. As most of the target area is urban type, we will use the urban insole to begin the deployment and conduct the first simulations and assessments. The template urban uses hexagonal cells, with 550 m cell radius and a single carrier. We deployment of Node Bs throughout the target area, the result is shown in Figure 18:
Figure 18: Deployment design and hexagonal cell radius 550 m. With an array of these features can cover the city's urban core with 36 locations. We will perform a first simulation to gauge whether the cell size and number of carriers is adequate or not. Atoll UMTS simulations are based on a Monte Carlo simulator [1].Since the user distributions of traffic map Atoll generates a population of users on the map and for each of these users the simulator executes a power control algorithm for the uplink and downlink.The objective of the algorithm is to minimize interference and maximize network capacity.This will restrict the connection to the network users who use low-priority services and generate a lot of interference.This process creates a snapshot of the UMTS network, the result is a distribution of users with different network parameters: level of interference, the terminal state (connected, connection refused ...), load factor for each cell, etc. In UMTS each mobile station receives interference from base stations other than their own cells, but not other phones, and all base station receives interference from their cell phones and other cells, but not the other base stations.
We have already said that UMTS capacity depends on the total received interference. Atoll simulates the power control mechanism using an iterative algorithm in each iteration, all the population of mobile users generated try to be connected, one by one, to the network. If certain users penalize others too mobile, they are rejected, with the decision of rejection correlated with service priority.In Atoll distinguished the following reasons for rejection: a) The signal quality is poor: · The carrier / interference in the DL is below the threshold (Ec / Io PTCH max). · Exceeding the maximum power that can transmit moving in the UL (Pmob> Pmob max). b) If the above restrictions are observed, the rejections are caused by network congestion: ·
It exceeds the load factor (in admission or congestion).
·
Have been exhausted channel elements per site.
·
Not enough power to transmit cell.
·
Have exhausted the spreading code.
A portion of the transmitter power is intended to pilot channel, another to the synchronization channel, another to control channels and the rest is shared among the traffic channels. Unlike the pilot channel and synchronization and control channels, the number of traffic channels and their powers depend on the data traffic, and is one of the parameters in the simulations is determined through the control algorithm power. The minimum and maximum power of traffic channels for each service are detailed in Table Services for UMTS Parameters.The sum of the power of traffic channels, control, synchronization, and pilot can not exceed the maximum power transmitted per cell. Instead of sticking to the results of a single simulation, we will perform a group of several simulations and study the results statistically. By running 10 simulations with all restrictions and value the results of the simulation average. The results obtained (on average) are shown in Tables 20-22 (in parentheses indicates the standard deviation): Traffic requested: Users
Active on the DL
Active in the UL
Active DL+ UL
Inactive
Total
3.684,8(68.6)
Voice
1.483,8
846
461.4
893.6
2.480,6(57.85) 595.3
593.6
398.1
893.6
MMS
136.8(8.28)
69.2
0
0
Internet access
1.005,6 (18.17) 820.9
183.2
1.5
0
Video Conference
61.8(9.04)
0
61.8
0
67.6
0
Table 20: Traffic demand at a given instant.
Simulation results (16.5 iterations on average per simulation): Number of users rejected
1867.9 (50.7%)
Exceeding the maximum power of the terminal in the UL (Pmob> Pmob max)
1.2
It exceeds the standard maximum power available for traffic Channels in DL (PTCH> PTCH max)
134.9
The carrier-interference in the pilot channel (DL) is below threshold (Ec / Io
1086
Saturation loading in the
635.6 DL
Refusal of admission
10.2
Table 21: Breakdown of rejected connections as the cause of rejection.
Broken down by services, Users
online
online
online
online
on the DL
in the DL
DL+UL
Services Total
1816(49.3%)(44.4)
459.2
448.4
300.6
Voice
1689.5(68.1%)(50.55)
398.1
413.9
268.8
MMS
17.1(12.5%)(4.93)
7.5
9.6
0
Internet access
78.7(7.8%)(8.94)
53.6
24.9
0.2
Video Conference
31.6(51.1%)(6.76)
0
0
31.6
Table 22: Breakdown of courses by the service connections
We can also study these models in a more graphic. Figure 19 shows the position of all the terminals at the time of the simulation are trying to access a service and the state found. In this case we see those red and black line that are being rejected or delayed.
Connection Rejection
Figure 19: Snapshot of the state of the network terminals. Visually, the results are consistent with the tables drawn from the simulations, we can see that about half of the users are being rejected. The simulation results are far from the established quality objectives.We see that indeed most penalized services are the lowest priority (MMS and Internet access) and more specifically the penalty is Internet access, which is what generates more interference. As the service requires the highest date rate, is the most traffic demand and therefore more traffic channels required and the cell that needs more power (generating interference in other phones).
Looking at Table 18, we see that the second cause of rejection is the saturation on the DL. That is, we do not have sufficient traffic channels to meet demand. In principle, the easiest way to increase the number of traffic channels is adding new carriers. And adding more transmitters also helps that there is more power to distribute among the traffic channels and may help to improve the quality of the signal, which would also be attacking the main cause of rejection (the carrier interference pilot channel (DL) is below the threshold (E c / I o
3.700,1 (43,56) 2.467,3 (54,02) 131 (12,03) 1.041,3 (14.86) 60.5 (7.76)
Active
Active
Active
In the DL 1.493,2
in the DL 864.2
DL+ UL 454.1
888.6
592.6
593.4
392.7
888.6
65.4
65.6
0
0
835.2
205.2
0.9
0
0
0
60.5
0
Table 23: Demand for a given traffic.
Simulation results (14.7 iterations on average per simulation): Number rejected (20.5%) Exceeding the maximum power of the terminal in the UL (Pmob> Pmob max) It exceeds the standard maximum power available for traffic channels in DL (PTCH> PTCH max) The carrier-interference in the pilot channel (DL) is below the threshold (Ec / Io
756.7 users 1.4 106.2 161.1 487.9 0.1
Table 24: Breakdown of rejected connections as the cause of rejection.
Broken down by services: Users
online
online In the DL
online in the DL
online DL+ UL
Services Total Voice MMS Internet Access Video Conference
2943.4(79.5%)(53.1) 2393.3(97%)(58.71) 75.3(57.5%)(9.49) 416.1 (40%)(9.61) 58.7(97%)(7.79)
905.7 574.7 36.4 294.6 0
735.7 575.5 38.9 121.3 0
440.6 381.7 0 0.2 58.7
Table 25: Breakdown of courses by the service connections. In this case we can represent the map of Seville on the results of these simulations (Figure 20).
Connection Rejection Figure 20: State of the terminal cells of 550 m radius and 3 carriers.
In this case shown on the map in Figure 20 the result of several simulations simultaneously. We see that the connection terminals are clearly more numerous, but the rejection rate remains high.
The results are greatly improved but still inadequate, we must rule out possible to cover UMTS to Seville with the cell size. Let's try using the following template available for deployment of UMTS Atoll in areas with high population density. The template dense urban target area divided into cells of 350 m radius, the result of covering the urban area of Seville with cells of this size would be the one shown in Figure 21:
Figure 21: Deploying UMTS cells 350 m radius. In this case the number of sites has increased significantly to 82. Initially we will size the network to its maximum capacity, ie, with three carriers per cell. If we find that the cell size is sufficient we can begin to reduce the number of carriers at less charged cells, to give him room for network growth and lower the initial cost of deployment. The simulation results are shown in Tables 26-28: Traffic requested: Users Total
3.669,9
Assets
Assets
Assets in
in DL
in UL
DL + UL
1.471,2
847
462,3
Inactive 889,4
(38,96) Voice
2.476,2
603,2
584,1
399,5
889,4
67,5
67,3
0
0
800,5
195,6
1,1
0
0
0
61,7
0
(41,11) MMS
134,8 (10,04)
Access
997,2
The Internet (18,82) Video
61,7
Conference (5,06) Table 26: Demand for a given traffic.
Simulation results (18 iterations on average per simulation): Number rejected users Exceeding the maximum power terminal in the UL (Pmob> Pmob max) It exceeds the standard maximum power available for traffic channels in the DL (PTCH> PTCH max) The carrier-interference pilot channel (DL) is below the threshold (Ec / Io
263.3 (7.2%) 0 30 3.1 230.2 0
Table 27: Breakdown of rejected connections as the cause of rejection.Broken down by services: Users
online
online On the DL
online in the DL
online DL+UL
Services Total Voice MMS Internet Access Video Conference
3406.6(92.8%)(46.39) 2474.9(99.9%)(41.05) 120.2(89.2%)(10.14) 750.1(75.2%)(15.31) 61.4(99.5%)(5.12)
1.239,5 603 59.4 577.1 0
816.5 583.6 60.8 172.1 0
461.7 399.4 0 0.9 61.4
Table 28: Breakdown of courses by the service connections.
In this case we can represent the map of Seville on the results of these simulations (Table 23):
Connection Rejection Figure 22: State of the terminal cells of 350 m radius and 3 carriers.
The results are still not achieving the quality objectives, so let's try to reduce a little the size of the cell. Predefined templates Atoll UMTS cell sizes do not allow minors. This is explained we've made a pretty optimistic traffic modeling (from the point of view of the operator) to cover our backs and make sure that the network later on staying small. We will define a template image of the dense urban, but with a cell size of 200 m 3 carriers. After making the deployment on the map the result is shown in Figure 22:
Figure 23: Deployment of UMTS cells of 200 m radius.
The simulation results shown in Tables 29-31: Traffic requested:
Total
Users
Claims on the DL
Active in the UL
Active in the Inactive DL + UL
3.684
1.488,33
830,33
470,67
894,67
602,33
583,67
415
889,4
62,33
63
0
0
823,67
183,67
1
0
0
0
54,67
0
(57,35) Voice
2.495,67 (16,65)
MMS
125,33 (13,72)
Access
1.008,33
The Internet (29,69) Video
54,67
Conference
(4,78)
Table 29: Demand for a given traffic.
Simulation results (17.33 average per simulation iterations): Number of users rejected Exceeding the maximum power terminal in the UL (P mob> P mob max) It exceeds the standard maximum available power for traffic channels in the DL (P tch> tch P max) The carrier-interference in the pilot channel (DL) is below the threshold (E c / I o
97 (2.6%) 0 13.33 0 83.67 0
Table 30: Breakdown of rejected connections as the cause of rejection.
Broken down by services: Users
online
online In the DL
online in the DL
online DL+ UL
Services Total Voice MMS Internet access Video Conference
3587(97.4%)(43.18) 2495.67(100%)(16.65) 122.33(97.6%)(13.72) 914.33(90.7%)(15.37) 54.67(100%)(4.78)
1.398,67 602.33 61 735.33 0
823 583.67 61.33 178 0
470.67 415 0 1 54.67
Table 31: Breakdown of courses by the service connections.
In this case we can represent the map of Seville on the results of these simulations. The state of the network shown in Figure 24:
Connection Rejection
Figure 24: State of the terminal cells of 200 m radius. As expected, virtually all of the requested connections have been accepted. These results if they meet the quality objectives set initially, we even have some room to try to minimize the cost of the network (number of sites) and reduce the number of carriers in some transmitters to provide a network for further margin expansion. To obtain these results are needed 250 locations.However, many of them are on the edges of the target area and only use 50% of the surface of some of their cells. It is expected that these sites are not providing service to many users and the traffic of these users can be taken up by neighboring cells without the degree of saturation increased significantly. Similarly, there are areas of the map with a density of users / traffic much smaller, so small cells do not need to support this traffic.200 m cells are essential in urban, dense urban buildings and in fact, in previous simulations most of the rejected users come from these areas. We rely on that to assume that if we remove the border sites to cover small target area and eliminate some sites open and industrial areas, the probability of
rejection need not be accepted.We dimensioned the network to the rejection in urban areas is acceptable, but in doing so we have oversized the network in other areas. After you delete and add sites several times and repeat the simulations as often as necessary were obtained the configuration of Figure 25:
Figure 25: Deployment of final locations of the UMTS network. We see that we have eliminated most of the sites on the edge of the target area and those in which only one cell was missed we reoriented the antenna to give coverage within the area of interest. We have also eliminated some sites of the environments with lower traffic density (open area and industrial area). In residential and industrial areas that are surrounded by dense areas and have maintained those sites that serve as reinforcement to support the traffic of the surrounding areas. We have also reoriented the antenna sites within the village that gave coverage to low traffic areas (such as parks, open type), to strengthen coverage of the surrounding areas more densely populated.The final configuration results are shown in Tables 32-34: Assets Assets Assets in Inactive Users in DL in UL DL + UL Total 3.647,25 1.460,75 898 841,75 446,75 (32,85) Voice
2.463,25
577,75
601
386,5
898
(34,37) MMS
122,5
64,75
57,75
0
0
818,25
183
2,75
0
0
0
57,5
0
(12,09) Access
1,004
The Internet
12.1
Video
57,5
Conference
(8,08)
Table 32: Demand for a given traffic.
Simulation results (16.75 average per simulation iterations): Number of users rejected Exceeding the maximum power terminal in the UL (P mob> P mob max) It exceeds the standard maximum available power for channels Traffic on the DL (P tch> tch P max) The carrier-interference in the pilot channel (DL) is below the threshold (E c / I o
98 (2.7%) 0 13.25 0 84.75 0
Table 33: Breakdown of rejected connections as the cause of rejection. Broken down by services: Users
online
online In the DL
online in the DL
online DL+UL
Services Total Voice MMS Internet access Video Conference
3587(97.4%)(43.18) 2463.25(100%)(34.37) 120.25(98.2%)(11.37) 908.25(90.5%)(4.66) 57.5(100%)(8.08)
1.398,67 577.75 63 729 0
823 601 57.25 177 0
470.67 386.5 0 2.25 57.5
Table 34: Breakdown of courses by the service connections. Reducing the number of sites by approximately 25% have achieved similar or even better for some services (MMS). This shows that some of the projected sites added nothing to the network and that something as simple as redirecting some antennas to areas of high traffic density can increase the network capacity. This is also confirmed in Figure 26, showing where the terminals are located rejected the previous simulations. We see that areas with higher density of sites are still the highest density of terminals has rejected, while the industrial area just west of the river has a dozen rejections with only 7 base stations.
Connection Rejection Figure 26: Location of the connections rejected.
3.6 Establishment of neighborhoods. Atoll is possible to establish automatically neighborhoods by imposing some restrictions on certain cells that may be part of a neighborhood. Once established neighborhood relations, Atoll easy viewing of neighboring cells on the map, which allows easy management. The algorithm for automatic assignment of neighboring cells is based on the following parameters: · -Max neighboring cells.It can be set globally or individually in the table cells. · Inter-Site-Max distance is the maximum distance that can exist between the reference cell and a cell candidate neighbor. · -Overlap between the coverage areas of the reference cell and a cell candidate neighbor.The concept of coverage here refers to the level of the pilot channel, or its signal to interference (Ec / Io).
·
-Power which contributes to the total interference.
Additionally you can set the following additional restrictions: ·
"Forcing all cells of the same site are neighbors.
·
-Force that are geographically adjacent neighboring cells.
·
Forcing symmetry-neighborly relations.
·
-Establish exceptional couples.
To perform automatic assignment of neighboring cells we will Automaticac allocation option, which is in Neighbours option within the cells of the folder option transmitters of the data tab of the browser window. We will impose the following restrictions on the establishment of neighborhood algorithm: Distance between neighboring sites: 1,200 m: in urban areas the distance between adjacent sites is around 600 m.But sites that cover open or industrial areas are more isolated, more than about 1,000 of the closest locations. This restriction aims at limiting the number of residents in such locations.Setting a maximum distance of 1,200 meters to ensure that these are neighboring sites only closer. Maximum number of neighbors: 20: This restriction is intended for sites in the most populated areas.Each cell is surrounded by a maximum of 6 other cells.If each physical cell Atoll are 3-cell (cell = torque transmitter / carrier) we will have 6 x 3 = 18 + 2 (the others carry the same physical cell) = 20. With this restriction we make sure that even if more than 20 pairs of transmitter / carrier that are less than 1,200 meters these are not considered neighbors. Atoll generates a huge table with all the neighbors of each cell. As such information becomes unmanageable will be included in Table 36, which shows only few cells have a given number of neighbors to see which is approximately the average number of neighbors per cell Number of neighbors 13 January 2005
Number of cells 6 9
10
26
9
13
8
54
7
137
6
1032
5
205
4
91
3
17
2
3
Table 35: Number of cells with a given number of neighbors.
3.7 Allocation of primary scrambling codes. The randomization codes allow you to separate from other cells. It is advisable to assign different codes to a given cell and all cells belonging to its list of neighbors. The assignment can be done manually for each cell, or automatically on all cells or a group of cells. Depending on the allocation strategy may be imposed various restrictions on code groups and domains, defining exceptional couples, distances and neighborhoods. At all times you can check the consistency of the current code assignment on the network under study. In UMTS there are 512 scrambling codes that are distributed in 64 clusters of 8 codes. The clusters are numbered from 0 to 63, and codes from 0 to 511.The code assignment can be done either manually or automatically. In the second case, Atoll provides a mapping tool based on an algorithm that takes into account the definition of groups and code domains, as well as additional restrictions based on the list of neighboring cells, second neighboring cells, criteria and minimum distance pairs exceptional. First let's create a code for domain Atoll. In the browser window, we will Transmitters | Cells | Primary Scrambling Codes | Domains and call codes Sevilla. It is essential that a cell and its neighbor does not have the same code as the maximum number of neighbors that we introduced in the calculation algorithm neighborhoods is 20 going to try initially to run the algorithm with 20 codes to see what it gives. In addition, we have seen that 20 is the number of cells in the space adjacent to another cell in the area of highest density of sites, so it seems a reasonable value to start. Table 36 lists the 20 codes are initially elected, in 4 groups of 5. Groups Group 1
Minimum 0
Maximum 4
Step 1
Group 2
32
36
1
Group 3+
64
68
1
Group 4
96
100%
1
Table 36: Codes of randomization initially elected. Before running the algorithm we have to go to the table cell (Cells Transmitters | Open Table) and fill the field Scrambling code domain with the domain created: Codes Sevilla. We can run the allocation algorithm. This option is in the browser window, in Transmitters | Cells | Primary Scrambling Codes | Automatic allocation.The associated dialog box, you can select the parameters that the algorithm takes into consideration: · -Existing Neighbours: using the table of neighborhoods, a cell can not have the same scrambling code to its neighboring cells, and between all codes must be different. · -Second Neighbours: the previous condition spreads to neighboring cells to their neighbors. · -Additional Ec / Io conditions: all stations belonging to the active set of the reference cell in the area where it provides the best signal, they must have different codes. ·
"Reuse Distance: Minimum distance from which codes can be reused.
We follow the same criteria to choose the number of codes. If forced to use different code to the neighbors of neighbors to 20 codes was too weak. It does not seem advisable to abuse of the codes that way in such a large network. Activate only as constraints Neighbours and Additional Existing Ec / Io conditions. On the other hand, we have seen that there is distance between neighboring sites if up to 1,200 m. Since it is very critical that we have in our network signals with power levels of the same order of magnitude using the same code in the same cell, we will be very restrictive in this regard and set a manifestly greater reuse distance: 2,000 m. The first execution of the algorithm given error, it was impossible to enforce these restrictions by using only 20 codes.So groups of 5 were added to the domain code of codes until the algorithm converged to reach a total of 55 codes. The codes used are those in Table 37. Groups
Minimum
Maximum
Step
Group 1
0
4
1
Group 2
32
36
1
Group 3+
64
68
1
Group 4
96
100%
1
Group 5
128
132
1
Group 6
160
164
1
Group 7
192
196
1
Group 8
224
228
1
Group 9
256
260
1
Group 10
288
292
1
Group 11
320
324
1
Table 37: Codes of randomization used. With the results of the algorithm, Atoll generates statistics that show the number of times the algorithm has assigned a specific code (Figure 27) and the number of times you have used a code for a given cluster (Figure 28 .)
Figure 27: Allocation of randomization codes for UMTS network. We have 177 sites, each with 3 sectors and 3 carriers per sector, which makes a total of 1,593 cells, for which we assigned 55 codes, which gives an average of
28.96 cells per code. We see that the code assignment revolves around the aforementioned value, indeed of the 11 sets of codes we see that there are 6 in all codes exceeded that average and 5 in which none does. Let us now use the cluster:
Figure 28: Use of cluster codes for our UMTS network. UMTS has 64 clusters of 8 codes each. We have defined a code domain consisting of 11 groups of 5 codes, each group therefore a distinct cluster. We then used 11 clusters, each with 3 free codes. Could therefore be added to the network 33 new codes without the need for a new cluster
3.8 Study coverage In this section we will perform a series of studies on the deployed network coverage. The aim of these studies is to document graphically the network and verify that the design is adequate. Coverage studies provide us with information on the status of the network at all locations of the target area. The different types of site surveys that can be performed in Atoll are: ·
Study coverage signal level.
·
Study transmitter coverage.
·
Study overlap.
·
Study of noise on the DL.
·
Study of signal to interference in the pilot channel.
·
Study of the service area on the DL.
·
Study of the service area in the UL.
·
Study of effective service area.
·
Handover study.
See the results of different studies: Study coverage signal level. The study provides a graphical representation of the signal level received by the terminal (downlink coverage.) The site survey performed by the signal level shown in Figure 30.
Figure 30: Study level of signal coverage. We have already said that UMTS is an interference limited radio system, so that the signal level who is not in principle limited coverage. In any case we have a signal level of -90 dBm over the entire target area (including interiors). Taking a value of a typical sensitivity of -105 dBm mobile terminal we have 15 dB of gross margin for fading, so in principle, the signal level should not be a problem for our network. Study transmitter coverage. This study will cover a different color mark the footprint of each transmitter in this case we used 10 colors and have been alternating for treating adjacent transmitters that do not match the same color. The result of the study are shown in Figure 31.
Figure 31: Study transmitter coverage. We see the sites in the village, the coverage area roughly coincides with the corresponding cell, while more isolated sites provide coverage to some areas significantly higher. Study overlap. This study shows the number of base stations that each point on the map above the threshold power at the reception.
The results of the study conducted overlap shown in Figure 32:
Figure 32: Study overlap. Study the level of interference in the DL. This study evaluates the total interference received in the downlink. All types of predictions that we will henceforth always refer to a simulation or set of simulations is performed for a terminal, service and mobility determined. We will conduct two studies on the level of interference, one for the majority case, telephone and voice terminal and one for the most critical case: terminal telephone and Internet access. These studies are done to the population generated by the simulation average of 10 made for the final configuration of the above.The study for the voice service and telephone terminal shown in Figure 33.
Figure 33: Study of the noise level of the Voice of the terminal telephone service. We see that the total interference level generated is significantly higher than the signal level of the site survey in Figure 30.Anyway it should not worry, because as explained in Chapter 2, UMTS systems are resistant to interference and due to the CDMA technology is easy to discriminate between the receptor interference and the desired signal. As discussed in the study of signal-interference in the pilot channel, to overcome a certain threshold of Ec / Io (UMTS typical value is 14 dB) is sufficient to discriminate signal and interference, and as will be seen in Figures 35 and 36 this occurs for almost the entire target area. The same goes for the Internet access service, but in this case the noise levels are even higher. This is because Internet access service requires a higher bandwidth and therefore needs more power transmission in the downlink. The results of the study for this service are shown in Figure 34.
Figure 34: Noise Study service Internet access to your terminal. Study of signal to interference in the pilot channel. This study places a test terminal type selected in each pixel and analyzes the relationship between E C / I O of the received signals.As in the previous case we have chosen the population generated by the simulation average of 10 simulations. Just as before, we will conduct a study to the most common (and your voice) and for the most critical case (telephone and Internet access). The results of the study for voice service and telephone terminal are the 35Figura
Figure 35: Ec / Io in the pilot channel for voice service telephone terminal Table 9 is set threshold E C / I O for all mobilities in -14 dB. We see that virtually the entire target area will have values above -15 dB, so in principle confirms the results of the simulations and we should not just rejection by poor signal quality for this service. We can see the results of the study to the Internet access service in Figure 36:
Figure 36: Ec / Io in the pilot channel for Internet access service telephone terminal. The results are very similar to the voice service, which is logical since the transmission power in the pilot channel is the same and the interference is the same for all services.The findings are equivalent to the previous case, we have an Ec / Io above -15 dB throughout the target area, so it is likely that there just rejections due to poor signal quality, confirming the results of simulations. Study of the service area on the DL. This study evaluates whether the test terminal can obtain service in the downlink, taking into account the limited traffic capacity based or active bases. This study is very interesting because it is the mobile that checks are rejected because of network congestion.We know that the power intended for traffic channels depends on the amount of traffic that has to be handed-over, and if at some point we have to transmit more power than the maximum, then there is traffic that has to be rejected, for which a running Atoll power control algorithm
that determines how much power goes to each connection and power connections are not (are rejected). This study places a test terminal at each location of the target area and see if you can get service or according to the results of simulations. The results for the telephone and voice terminal are shown in Figure 37.
Figure 37: Study of the service area on the DL for the terminal voice phone service. We see that we can get service at all locations of the target area. This is consistent with the results of simulations and for the voice service had no rejections. We must remember that this does not mean never going to have rejections for this service. This means that in 10 (which are the times you have repeated the simulation) we have made snapshots of the network with traffic demand within the normal range, there were no rejections.
In exceptional situations, where demand for passenger traffic to grow, such as disasters, Fair, Easter, New Year ... it certainly will be significant even rejection rates for voice service. The results of the study for the terminal telephone and Internet access are shown in Figure 38.
Figure 38: Study of the service area on the DL for the Internet access service telephone terminal. We see that most of the target area can get service, but especially in dense urban areas, there are some locations where our connection attempts would be rejected.This confirms the results of the simulations, they gave us half a rejection rate of 9.5% for this service.Of course, the majority of these rejections would occur in the area of greatest density. Study of the service area in the UL. It is analogous to the above but for the uplink, taking into account the limited power of the mobile terminal.
The results of the study for the terminal telephone and voice are as shown in Figure 39.
Figure 39: Study of the service area in the UL for voice service telephone terminal In other mobile communications systems such as GSM or TETRA uplink is usually more limiting than the downward, as the need to take small, manageable terminals forces us to take power in the upstream transmission very low and not get compensated designing high sensitivity receivers in base stations. However, our UMTS network behaves the opposite. As seen in Figures 39 and 40 get service in the increase in all locations of the target area, which agrees with the results of the simulations, which gave us 0 rejections excess load on the ascendant. This is logical because the Internet access service has a very asymmetric traffic with a high demand in the downstream and far less on the up, it makes sense that the network has to reject many more connections than the other.
Figure 40: Study area in the UL service for Internet access service telephone terminal Study of effective service area. This study provides the area intersection of the two. As we have seen, the downlink is much more restrictive than the upwardly, so the results of this study are virtually identical to the study of the service area in the downlink. The results for the telephone and voice terminal are shown in Figures 41 and 42:
Figure 41: Study of effective service area to service your voice terminal
Figure 42: Study of effective service area to service Internet access to your terminal Handover study. This prediction studies the active set of a test mobile located at each point on the map, and renders it according to selected criteria.Let us briefly explain the concept of the active set in UMTS. In the UMTS system uses a handover mechanism for transferring called continuity, SHO (Soft / Softer Handover).Thanks to universal frequency reuse is possible to connect the call to the candidate to the handover station before disconnecting it from the source station, keeping both links simultaneously for some time. A call can be supported by the three sectors of a base station and / or by two or more stations. Each of the bases involved keeps in touch with the phone until the attenuation to one of them is excessive, when you leave the link on that basis. In the uplink, during the handover period continuously, the signal transmitted by the mobile is detected by the base stations involved, make a selection or combination of demodulated signals. In general, for base stations located at different sites, it is easier to select the signal of higher quality (soft
hand-off).For base stations located on the same site, as in sectorized cells, the physical proximity to combine the signals (soft hand-off) before demodulation. The set of bases with a mobile is known as the Joint Contact Active (Active Set).The maximum number of stations that can be part of the active set of a mobile (Active Set Size) depends on the type of terminal. The criteria used for a station is part of the active set of a terminal is based on the concept of threshold for handover (AS THRESHOLD), defined for each cell in the table Transmitters | Cells | Open Table.The transmitters that constitute the active set of a tower should meet the following conditions: ·
"They must use the same frequency
· "The quality of the pilot (Ec / Io) of the best season to exceed a threshold defined for terminal mobility (in this case -14 dB). · "The pilots of the other bases in the active set must have a Ec / Io that does not fall below the threshold of handover on the best season. · "They must be nearby stations of the best base if you selected AS Neighbours restricted to the characteristics of the equipment. The results of this study are shown in Figures 43 and 44:
Figure 43: Study of the asset to the Voice of the terminal telephone service Just as occurred in studies of Ec / Io, these studies are identical for both services and the level of the pilot channel signal and interference are the same for both. As before, studies have been done to the population generated by the average of 10 simulations of the final configuration.
Figure 44: Study of active Internet service's terminal access to your Study of interference in the pilot channel. At each pixel indicating whether the number of bases that are received Ec / Io enough is "excessive" in the sense that exceeds the maximum number of active bases allowed by the choice of mobile terminal. It is appropriate that each location on the network has the maximum number of active bases, as this benefits the soft handover, but if we have more bases than allowed only thing is to get more signal level waste, which translates into interference.Any signal that arrives at a terminal that is not one of its bases is active interference. Interference studies conducted in the pilot channel are shown in Figures 45 and 46
Figure 45: Study of interference in the pilot channel for the service of Voice of the telephone terminal. We hardly have seasons interfering in most locations. This affects very low levels of denial of connections to the network (as we have seen in the simulations, where we have not had any rejection for voice service).
Figure 46: Study of interference in the pilot channel for Internet access service telephone terminal. In this case the number of interfering base stations is much higher, which results in a rejection rate higher than for voice service (which we have seen in the simulations). This is because Internet access service is not configured to support soft handover, and therefore any base station exceeds the threshold Ec / Io instead of being part of the active set, it becomes an interfering base station.
3.9 Network evolution. As mentioned above, the network has been designed to meet quality objectives when the forecast of subscribers has reached its peak (20% of the inhabitants of the city).It is expected that the number of subscribers later years to reach those numbers, or you may not even reach these amounts. Have also discussed the difficulties involved for a CDMA system the handover between different frequencies, and that requires working in compressed mode, so they agree to limit as far as possible the areas that must be produced using various carriers. Not seem necessary or desirable then the use of three carriers per cell since the launch of the network. The logical thing is to make an initial deployment with a single carrier per cell into expanding capacity to measure the number of subscribers need them. In this section we make a study of how it degrades the quality of network service as the number of subscribers increases and substantial improvements are made as extensions of the building. Atoll can easily simulate such scenarios due to the scaling factor. When the simulation is allowed to use a parameter called Global Scaling Factor, which scales the traffic demand by multiplying by the scaling factor.A factor of 0.4 means that the simulations were performed considering 40% of actual traffic. Let us assume that the number of subscribers increases linearly at a rate of 10% maximum per year to reach the maximum number of subscribers to 10 years of the implementation of the network. As mentioned above, initially carried out the deployment with a single carrier per sector. 1 year after deployment (1 carrier per cell, 10% of subscribers): Online Services Total
345.7 (99.4%) (16.03)
Voice
234.4 (100%) (17.01)
MMS
12.9 (98.5%) (1.81)
Internet Access
93.3 (98.1%) (7.27)
Video Conference
5.1 (100%) (1.64)
Table 38: Connections accepted by the network (10% of users and 1 carrier). We found that indeed, there is no need to start the deployment with the racks at full capacity. This allows us to significantly reduce the initial investment for setting up the network and guarantees a better initial performance of the network (at work with only one carrier per sector). Avoid oversizing and also get the network itself to finance its expansion of capacity, since the network is operational and therefore billing from day one. Besides the capacity expansion of the network are zero-risk investment, because traffic is rejected and the money we lose the ability to increase income means increasing systematically. 2 years after deployment (1 carrier / cell, 20% of subscribers): Services
Online
Total
720.7 (98.6%) (23:43)
Voice
490.3 (100%) (17.81)
MMS
25.5 (99.6%) (6.76)
Internet Access
191 (94.9%) (18.07)
Video Conference
13.9 (100%) (2.07)
Table 39: Connections accepted by the network (20% of users and 1 carrier). Quality objectives continue to be met comfortably 2 years after completion of the deployment. In addition to reducing the investment required for implementation of the network, another advantage of not displaying all the carriers from the initial moment is that you avoid the wear suffered all these carriers to be operational, thus prolonging the life of the network and Minimizing the number of failures in the network (unless carriers, lower failure rate). This also saves on maintenance and improvement in the quality of network service. 3 years after deployment (1 carrier / cell, 30% of subscribers): Services
Online
Total
1079.5 (97.3%) (9.22)
Voice
745.7 (100%) (17.94)
MMS
41.7 (98.6%) (5.73)
Internet Access
272.9 (90.3%) (9.22)
Video Conference
19.2 (99.5%) (4.21)
Table 40: Connections accepted by the network (30% of users and 1 carrier). We see that in this case the quality objectives are met by a small margin. It is clear that during the 3 rd year of operation of the network will have to start to be the first expansion of capacity, adding a carrier in the most loaded (which will probably make time to be risen from 10% to reject the service Internet access, as these data provide for the entire network and would ideally be met for all cells). We will simplify and as we have done until now we demand that quality objectives are met in all cells, but only at the entire network (most operators do not or globally).We will not make the expansion of capacity in a progressive manner, which would be optimal. Let's go on pretending until the quality objectives are no longer met, at which will double the network capacity by adding a carrier to all sectors. 4 years after deployment (1 carrier / cell, 40% of subscribers): Services Total
Online 1408.9 (95.8%) (42.97)
Voice
990.2 (100%) (39.32)
MMS
49.8 (94.3%) (7.24)
Internet Access
343.8 (85.5%) (15.09)
Video Conferencing
25.1 (100%) (3.08)
Table 41: Connections accepted by the network (40% of users and 1 carrier). As expected, after 4 years of operation of the network of low priority services have fallen below the quality objectives.It is therefore the time of the first expansion of network capacity. Adding one carrier per sector the results of the simulations are: 4 years after deployment (2 carriers / cell, 40% of subscribers): Services Total
Online 1426.1 (98.8%) (20.44)
Voice
983.8 (100%) (37.02)
MMS
47.3 (99.4%) (6.42)
Internet Access
370.1 (95.8%) (20.44)
Video Conference
24.9 (99.6%) (5.79)
Table 42: Connections accepted by the network (40% of users and 2 carriers). The capacity expansion has had the expected and the network could grow a few more years before needing a new extension. 5 years after deployment (2 carriers / cell, 50% of subscribers): Services
Online
Total
1817.38 (98.4%) (37.06)
Voice
1259.5 (100%) (28.24)
MMS
63.63 (99%) (5.1)
Internet Access
464.38 (94.3%) (22.04)
Video Conference
29.88 (100%) (3.95)
Table 43: Connections accepted by the network (50% of users and 2 carriers). Quality objectives continue to be met satisfactorily. However, it is normal at 5 years is thought to migrate to new technology (in this case would be HSDPA). It is accepted that between the deployment of each generation of mobile spend about 10 years and that 5 is normal to migrate to an intermediate technology. In fact, we know that the first UMTS network in Seville began operation on June 1, 2002 and five years later, in 2007 and we cover HSDPA (3.5G) [10]. With this new technology, it is possible that the capacity expansion planning when otherwise it would be appropriate to consider that if the capacity expansion that migration brings is enough to meet increased traffic demand or need to continue to deploy carriers. Anyway, HSDPA escapes the objectives of this project, so it is going to plan the expansion of capacity without regard to migration. 6 years after deployment (2 carriers / cell, 60% of users): Services
Online
Total
2158.67 (97.7%) (45.55)
Voice
1472.17 (100%) (50.69)
MMS
84.67 (97.3%) (7.72)
Internet Access
564.33 (92%) (19.8)
Video Conference
37.5 (100%) (4.57)
Table 44: Connections accepted by the network (60% of users and 2 carriers). Already beginning to be seen again as the increase in the number of subscribers is gradually degrading the quality of service.However, following the same approach as before, not to simulate 3 carriers to fall below the quality objectives. 7 years after deployment (2 carriers / cell, 70% of users): Services
Online
Total
2498.6 (96.9%) (40.58)
Voice
1755.6 (100%) (45.26)
MMS
86 (96.6%) (7.27)
Internet Access
614.2 (88.9%) (10.53)
Video Conference
42.8 (100%) (5.74)
Table 45: Connections accepted by the network (70% of users and 2 carriers). We see that for some, but have fallen back below the quality objectives. It is therefore the time of the last upgrade of the capacity of our network. 7 years after deployment (3 carriers / cell, 70% of capacity): Services
Online
Total
2523.8 (98.6%) (65.56)
Voice
1728 (100%) (30.05)
MMS
96.6 (98.4%) (10.97)
Internet Access
654 (95.3%) (33.24)
Video Conference
45.2 (100%) (5.42)
Table 46: Connections accepted by the network (70% of users and 3 carriers). As we expected, with 70% of the number of subscribers and the network to its maximum capacity exceeded the targets. Little else is there to comment, Tables 39, 40 and 41 show the simulation results for 80%, 90% and 100% of the number of subscribers expected. 8 years after deployment (3 carriers / cell, 80% of users):
Services
Online
Total
2886.8 (98.3%) (54.88)
Voice
1972.2 (100%) (32.52)
MMS
103.6 (98.9%) (12.88)
Internet Access
761.6 (93.9%) (30.86)
Video Conference
49.4 (100%) (10.25)
Table 47: Connections accepted by the network (80% of users and 3 carriers). 9 years after deployment (3 carriers / cell, 90% of users) Services
Online
Total
3201.5 (97.9%) (52.41)
Voice
2187 (100%) (36.4)
MMS
119.75 (97.6%) (9.6)
Internet Access
847 (92.8%) (30.85)
Video Conference
47.75 (100%) (6.87)
Table 48: Connections accepted by the network (90% of users and 3 carriers). 10 years after deployment (3 carriers / cell, 100% of users): Services Total
Online 3587 (97.4%) (43.18)
Voice
2463.25 (100%) (34.37)
MMS
120.25 (98.2%) (11.37)
Internet Access
908.25 (90.5%) (4.66)
Video Conference
57.5 (100%) (8.08)
Table 49: Connections accepted by the network (100% of users and 3 carriers). This raises the question of what to do after 10 years when the number of subscribers continues to grow and the quality of network service getting worse, "increase the number of sites? The answer is clearly no. Even the most visionary could guess back in 2000 when these networks were planned just as GPRS (2.5 G) served as a bridge between GSM (2G) and UMTS (3G) technologies appear to provide increased capacity of
UMTS, HSDPA and HSPA already in operation, 5 years after the launch of UMTS in Seville and there is talk that it is possible the emergence of the first 4G network in the United States later this year [10]. The evolution of technology is virtually unpredictable and as I said before is not thought advisable to design networks that will be in operation for many years may become obsolete before starting to recover its investment. 4 G may begin to appear at any time between 2008 and 2012. This means that a UMTS network in Seville may have a lifespan of well over 10 years, and if it turns out, are always bridges to support technologies that increase in traffic demand.