Frequency Planning
Overview Page 2
FP Principle FP Definition Network evolution Cell Planning - Frequency Planning Interference Theory Carrier types - Multiple Reuse Pattern MRP Manual Frequency Planning BSIC Planning
Frequency Planning Process RMS Frequency Planning
Frequency Planning Principle
FP Definition Page 4
Frequency Planning is the process made o provide to each TRX in the network a certain frequency Frequency planning is done in order to provide the highest spectrum efficiency (higher capacity with less resources) The frequency allocation is done in such way that all the FP constraints are fulfilled: Co-cell/co-site/neighbour separation constraints Overall interference is reduced
GSM Frequency Spectrum Page 5
GSM 850 DL: 824-849 MHz, UL: 869-894 MHz 200 kHz channel spacing 124 channels
ARFCN 128 -251
GSM 900 DL: 935-960 MHz, UL: 890-915 MHz 200 kHz channel spacing 124 channels
ARFCN 1 – 124
E-GSM DL: 925-935 MHz, UL: 880-890 MHz 200 kHz channel spacing Additional 50 channels
ARFCN 0, 975 - 1023
GSM 1800 DL: 1805-1880 MHz, UL: 1710-1785 MHz
200 kHz channel spacing 374 channels
ARFCN 512 – 885
GSM 1900 DL: 1850-1910 MHz, UL: 1930-1990 MHz
200 kHz channel spacing 300 channels
ARFCN 512 -810
Impact of limited Frequency Spectrum Page 6
Bandwidth is an expensive resource Best usage necessary Efficient planning necessary to contain good QoS when the traffic in the network is increasing smaller reuse (high traffic with less resources) MRP usage implementation of concentric cells / microcells/dual band implementation of Frequency Hopping – Baseband Hopping (BBH) – Synthesized Hopping (SFH or RFH)
Network Evolution - Coverage Approach Page 7
First issue in a network life-cycle is to provide coverage Network design changes rapidly The frequency plan has to be adapted after each network extension Planning method must be flexible and fast (group method) During first steps manual frequency planning possible
Network Evolution - Capacity Approach 1/3 Page 8
With the growing amount of subscribers, the need for more installed capacity is rising Possible Solutions: Installing more TRXs on the existing BTS Implementing additional sites
More frequencies on air higher interference
Network Evolution - Capacity Approach 2/3 Page 9
Installing more TRXs - Advantages No site search/acquisition process No additional sites to rent (saves cost) Trunking efficiency Higher capacity per cell
Installing more TRXs - Disadvantages More antennas on roof top (Air combining) Additional losses if WBC has to be used Less (indoor) coverage More frequencies per site needed Tighter reuse necessary decreasing quality/ increase interference
Network Evolution - Capacity Approach 3/3 Page 10
Implementing additional sites - Advantages Reuse can remain the same (smaller cell sizes) Needs less frequency spectrum higher spectrum efficiency
Implementing additional sites - Disadvantages Site search/acquisition process needed Additional site cost (rent) Re-design of old cells necessary (often not done)
What is frequency reuse? Page 11
As the GSM spectrum is limited, frequencies have to be reused to provide enough capacity The more often a frequency is reused within a certain amount of cells, the smaller the frequency reuse Aim: Minimizing the frequency reuse for providing more capacity REUSE CLUSTER: Area including cells which do not reuse the same frequency (or frequency group)
RCS and ARCS 1/2 Page 12
Reuse Cluster Size - RCS If all cells within the reuse cluster have the same amount of TRXs, the reuse per TRX layer can be calculated: B
RCS
# TRX / cell
Average Reuse Cluster Size - ARCS If the cells are different equiped, the average number of TRXs has to be used for calculating the average reuse cluster size: ARCS
B
# TRX / cell
RCS and ARCS 2/2 Page 13
The ARCS is giving the average reuse of the network when using the whole bandwidth and all TRXs per cell E.g: if we want to have the reuse of all non hopping TCH TRXs, we have to use the dedicated bandwidth and the average number of non hopping TCH TRXs per cell to get the ARCS of this layer type. Each cell has only one BCCH. Therefore the BCCH reuse is an RCS and not an ARCS! The lower the ARCS is: the higher is capacity traffic (more TRX/cell) The higher interference is
Traffic capacity/inference is always a trade-off
Reuse Cluster Size 1/2 Page 14
Sectorized sites 4 sites per reuse cluster 3 cells per site REUSE Cluster Size: 4X3 =12
1
2
4
3 1
2
4
3
6 7
5
8 9
8 9
6 7
10
11 12
5
10
11 12
Reuse Cluster Size 2/2 Page 15
Sectorized sites 3 sites per reuse cluster 3 cells per site 1
REUSE Cluster Size 3X3 = 9
2
4
3
5 6
7
8 9
1
2
4
3
5 6
7
8 9
Frequency Reuse: Example Page 16
No sectorization
BCCH RCS
7 cells per cluster BCCH RCS = 7
TCH Reuse: Depending on BW and Number of installed TRXs per cell
interferer region
Example:
TCH RCS
B= 26 4TRXs per cell TCH RCS
26 7 BCCH 1Guard 3
6
Cell Planning - Frequency Planning Page 17
Can frequency planning be seen independently from cell planning?
Discussion Bad cell planning Island coverage
disturbing the reuse pattern
Big overlap areas
bigger reuse necessary
Good cell planning Sharp cell borders
good containment of frequency
Small overlap areas
tighter reuse possible
Influencing Factors on Frequency Reuse Page 18
Topography Hilly terrain Usage of natural obstacles to define sharp cell borders tighter frequency reuse possible Flat terrain Achievable reuse much more dependent on the accurate cell design
Morphology Water
low attenuation
high reuse distance
City
high attenuation
low reuse distance
Examples for different frequency reuses Page 19
Big city in the south of Africa: BCCH reuse 26 Irregular cell design Mixed morphology Lots of water Flat terrain plus some high sites
Big city in eastern Europe BCCH reuse 12 Regular cell design Flat area Only urban environment
Interference Theory 1/2 Page 20
C/I restrictions 9dB for co-channel interference -9 dB for adjacent channel interference P rec
Received Power Prec, A
Prec, B
C/ I
0
R
dista nce
D
Interference Theory 2/2 Page 21
ARCS 6.5..9.0 7.0..9.5 8.5..11.0 12.0..16.0
Interferer probability C/Imed is the calculated carrier to interferer ratio at a certain location (pixel) Interferer
Probability density function [%]
4,0%
80 %
3,0%
60 %
2,0%
40 % Margin
20 %
0,0% C/Ithr
%
100%
5,0%
1,0%
robabilit
Pint[%] 10 7.5 5.0 2.5
C/Imed
C/I [dB]
0%
-20
-15
-10
-5
0
5
10
15
C/I - C/Ith r [dB]
20
Carrier Types - BCCH carrier Page 22
BCCH frequency is on air all the time with full power BCCH BCCH interference is always present
If there is no traffic/signalling on TS 1 to 7 dummy bursts are transmitted No interference reductions mechanisms are allowed, like: PC (Power Control) DTX (Discontinuous Transmission) are not allowed
BCCH needs a clean frequency plan since it is used for all mobiles within the network
Carrier Types - TCH carrier Page 23
PC allowed and recommended for UL and DL Reduction of transmit power according to the actual path loss Careful parameter tuning for DL necessary
DTX allowed and recommended for UL and DL Discontinuous Transmission If there is no speech, nothing is transmitted Generation of comfort noise at receiving mobile
TCH not in use no signal is transmitted Special case: Concentric cells Different re-uses for inner and outer zone are possible
Multiple reuse pattern 1/2 Page 24
For different types of carriers, different interference potential is expected BCCH layer needs a higher REUSE then on other layers: as the BCCH carrier has the highest interferer potential because of being on air all the time the BCCH channel itself is accepting only low interference
TCH layers can be planned with a smaller REUSE Inner zones of concentric cells are able to deal with the smallest reuse in non hopping networks
Multiple reuse pattern 2/2 Page 25
REUSE clusters for
INNER ZONE layer
TCH layer
BCCH layer
GSM restrictions Page 26
Intra site minimum channel spacing 2 Intra cell minimum channel spacing 2 from Alcatel G2 BTS, (3 specified by GSM standard) . . . ,
3 A
f ,
2
A f , 1 A
f
Frequencies fAx,fBx,fCx,… must have at least 2 channels spacing Frequencies fx1,fx2,fx3,… must have at least 2 (or 3 depending on HW used) channels spacing
Intermodulation problems 1/2 Page 27
IM Products GSM900 In a GSM 900 system intermodulation products of 3rd and 5th order can cause interference 2 * f 1,t – f 2,t = f 2,r / 2 * f 2,t – f 1,t = f 1,r 3 * f 1,t – 2 * f 2,t = f 2,r / 3 * f 2,t – 2 * f 1,t = f 1,r
Frequency planning must avoid fulfilling these equations Both frequencies must be on the same duplexer To avoid intraband IM inside GSM900 the following frequency separations shall be avoided: 75/112/113 channels IM5
IM3
Intermodulation problems 2/2 Page 28
IM Products GSM1800 In a GSM 1800 system, only intermodulation products of 3rd order can cause mesasurable interference 2 * f 1,t – f 2,t = f 2,r / 2 * f 2,t – f 1,t = f 1,r Frequency separations to be avoided 237/238 channels
IM Products Dual Band (GSM900/GSM1800) f 1800,t – f 900,t = f 900,r Decoupling between the GSM 1800 TX path and the GSM 900 RX path is less than 30 dB (e.g. same antenna used!)
Treating “neighbour” cells Page 29
Cells, which are not declared as neighbour cells but are located in the neighbourhood may use adjacent frequencies if it is not avoidable, but no co channel frequencies Cells which are declared as neighbours, thus have HO relationships, must not use co or adjacent frequencies If an adjacent frequency is used, the HO will be risky and at least audible by the user. Sometimes due to big frequency constraints separations of 1 channel for neighbour cells (with low amount of HO) is acceptable.
Where can I find neighbour cells? Page 30
At the OMC-R for each cell a list of neighbour cells is defined Maximum number of neighbours: 32 The list of neighbours and their frequencies is transmitted to the mobile to be able to perform measurements on these frequencies In case of a HO cause, the HO will be performed towards the best neighbour
BSIC
BSIC allocation Page 32
Together with the frequencies the Base Transeiver Station Identity Code (BSIC) has to be planned BSIC = NCC (3bits) + BCC (3bits) The BSIC is to distinguish between Base Stations using the same BCCH frequency The aim of BSIC planning is not to use the same frequency/BSIC combination on cells influencing each other BSIC can be planned by the A9155 RNP tool
Spurious RACH Page 33
Bad BSIC planning can cause SDCCH congestion cause by the spurious RACH problem, also known as “Ghost RACH” This problem occurs, when a mobile sends an HO access burst to a TRX of cell A using the same frequency as a nearby cell B uses on the BCCH Both cells using the same BSIC and Training Sequence Code TSQC, the HO access burst is understood by the cell B as a RACH for call setup Therefore on cell B SDCCHs are allocated everytime a HO access burst is sent from the mobile to the cell A
Frequency Planning Process
Network Life Cycle Page 35
Frequency planning occurs in all phase during network life-cycle During planning phase (roll-out phase)
During optimization phase
Frequency Planning Process - Contents Page 36
Analysis of existing FP FP Inputs FP Strategy Definition Preparation Work FP Creation Frequency Plan validation Frequency Plan Implementation Post Implementation Tasks Reporting
Analysis of existing FP
Analysis of existing FP 1/2 Page 38
This step is done during: Optimization phase Network densification steps
The reason is to define: Define if available RNP data can be used for the generation of a new FP Sites coordinates, cell impacted, height, azimuths
The strategy used for FP (non-hopping, hopping) The possible outcome of a new FP KPI expected to be improved Define if a FP is needed
Spectrum definition Possible coverage problem FP brings no benefit in these areas Hardware used define co-cell the channel separation (2 or 3)
Analysis of existing FP 2/2 Page 39
Analysis of existing FP consists in: A9155 analysis Import csv files into A9155 (through A9155 PRC Generator) Coverage plots areas with bad coverage C/I Plots areas with high interference
Separation constraints violation Analysis of the usage of the frequency band Define type of hopping OMC-R analysis Define areas with low KPI Analysis of the frequencies from that areas
Reporting Should be the base of the FP strategy chosen
FP Inputs
FP Inputs Page 41
FP Inputs must be provided by customer before starting the FP. target area of the new frequency plan has to be defined, as well as the list of
all involved cells from this area. frequency spectrum. The FP targets must contain the available frequencies. If there are usage constraints related to the frequency spectrum they must be provided. Day Y, when the new frequency plan has to be ready for implementation. The new plan must take into consideration the network configuration planned for this day. FP Quality Indicators used for frequency plan validation (before implementation) Not the well know KPI New indicators: like C/I plots
The expected results from the FP should be clearly stated from the beginning, and the whole strategy should be driven by these goals.
FP Strategy Definition
FP Strategy 1/3 Page 43
FP Strategy contains the different methods used during frequency allocation process It consists of defining: Spectrum Partitioning Macro layer / Micro layer BCCH / TCH Guard Bands / Joker Frequencies Decision on Frequency Hopping Implementation for QoS improvement due to capacity saturation
FP Strategy 2/3 Page 44
Frequency Coordination at the Planning/Country Border Planning Border Take into consideration the frequencies of the first ring outside planning area Country Border as defined in ETC recommendation
Frequency Coordination at Co-Existence of Several Systems. It must avoid Spurious Emissions Receiver Blocking Intermodulation Products
BSIC Allocation Strategy
FP Strategy 3/3 Page 45
Frequency Planning Activation Mode Message Mode (PRC activation) Massive Logical Update (MLU)
Definition of Hot Spot Areas Set a higher priority during FP for areas with high traffic
Handling of Sites with Untypical Configurations
FP Preparation Work
FP Preparation Work 1/2 Page 47
Retrieve data needed for FP: Logical data (csv files) Physical data: sites coordinates, heights, antenna types, tilt…
A9155 AFP Dry Run Identify the possible problems that might occur and the time needed
OMC Neighbors Relationships Clean-up Bad neighbour planning bad frequency plan A problem in most running networks too many neighbours declared Neighbour plan to be checked/optimized
FP Preparation Work 2/2 Page 48
Experience Database Computed from field feedback during network operation It is based on: Old reports Anomaly reports RNP/RNO experience
Prepare Before/After Comparison In order to compute the KPI before FP implementation (for a later comparison) Can be: Drive tests OMC-R Statistics
FP Creation
FP Creation Page 50
FP is created using a A9155 AFP Module BSIC planning has to be done as well All parameters defined during the strategy phase should be reflected in the tool (see AFP training)
Frequency Plan Validation
FP Validation Page 52
Validation is done to take the decision about the implementation of a new FP There are several means of evaluating a FP (before implementation) Interference calculation Constraints violation Visual analysis of frequencies plan Frequency distribution.
KPI cannot be used in this phase
Frequency Plan Implementation
Frequency Plan Implementation Page 54
Implementation of the frequency plan is done via OMC-R through the PRC PRC can be created: Manually For very small changes Using External Tools A9155 PRC Generator Module
Post Implementation Tasks
Post Implementation Tasks Page 56
To check the frequency plan after implementation, intensive QoS analysis must be performed OMC-R Drive tests
Compute KPI for before/after comparison The optimization solutions for possible problems are: Using joker frequencies Use MAFA to find “clean” frequencies Manual optimization
All problems occurred must be reflected in an updated Experience Database (used for future FP)
Reporting
Reporting Page 58
Reporting step should contain before/after comparison, comparison, to At least one week network monitoring for before/after show The QoS Improvement Capacity increasement increasement (if this was the main reason for FP) entire FP process process All problems encountered during entire improvements/suggestions in the overall Alcatel FP Process Possible improvements/suggestions
New FP method: RMS based FP
Radio Measurement Statistics Page 60
RMS creates statistics on the network QoS: based on field measuremen measurements ts
Measurements ts are performed by each mobile during a call Measuremen
RMS Different Outputs at TRX level on neighbor cells (C/I)
Used for RMS FP
on downlink and uplink quality & level on number of consecutive bad speech frames (BFI) on Radio Link Counter (UpLink only) on Path Balance and on Timing Advance on Power and number of channel seizures 9 Voice quality indicators
RMS Frequency Planning Basics Page 61
The Frequency Planning target is to improve networks QoS to reduce overall network interference
For all cells, RMS creates a C/I report for each neighbour Measured interference is used by A9155 for frequency planning RMS Measurement Neighbour C/I
RMS Measurements
Generate RMS files
OMC-R
A9155 Creates IM
A9155
New FP based on RMS IM
A9155
RMS Frequency Planning Page 62
RMS limitation for FP: Measurements are done only for declared neighbours in OMC-R. This leads to: Interferer BCCH frequencies, which are not neighbours, are not measured RMS interference matrix is not reflecting the complete network interference: – Report only for declared neighbours – Interference is existing but is not measured
Solution: Generate artificially more neighbours during RMS measurements (dummy neighbours) Dummy neighbours can be: Real cells (with very high HO_MARGIN) Logical cell with the BCCH frequency to be measured (dummy cells)
Dummy Neighbours Principle Page 63
RMS measurements without dummy neighbours Cell close to serving cell but not neighbours are: not reported not contributing to overall interference matrix Some interferer BCCH frequencies are not reported
Serving cell
Neighbour cell
Other cells
Dummy Neighbours Principle Page 64
RMS measurements with dummy neighbours All cells close to the serving cell are measured and reported to the OMC-R Target is to measure as many different BCCH as possible Dummy neighbours: Not real neighbours Different BCCH than: – Serving cell – Neighbour Cells
Serving cell
Neighbour cell
Dummy neighbour cell
RMS Based AFP Process Flow Page 65
Conclusions
Discussion: Subdivide Frequency Band? Page 67
Any subdivision of the frequency band is reducing the spectrum efficiency! As the BCCH has to be very clean, it is nevertheless recommended to use a separated band and select a bigger reuse Microcells/Inner zones of concentric cells do not need a separated band if they are dedicated for hotspot coverage If there is a continuous layer of microcells/inner zones, it makes planning easier when using a separated band Any other separations should be avoided if possible!
Hint for creating a future proofed FP Page 68
If a frequency plan is implemented, using all available frequencies in the most efficient way, it is very difficult to implement new sites in the future! New sites would make a complete re-planning of the surrounding area or the whole frequency plan necessary To avoid re-planning every time when introducing new sites, it is recommended to keep some Joker frequencies free These Joker frequencies can be used for new sites (especially BCCH TRXs) unless it is impossible to implement new sites without changing a big part of the frequency plan New frequency plan necessary!
