Mentum planet v5.3 LTE TDD case study for NSN Hangzhou trial
Peter Cheung, Technical Consultant Mentum HK 10 Jun 2011 (updated 15 Jun 2011)
summary Use planet v5.3 • Aim • Convert NSN Hangzhou LTE TDD trial project • Optimize site based on some KPI example
• Input • • • •
2D maps (clutter, height, clutter height) 3D maps (building polygon with AGL) NSN project data (site long/lat, PCI, freq, antenna, NL) Default Default site config (power, (power, loading) loading)
• 3D Propagation model • UM model at 2.6GHz with 2D + 3D maps • With default building penetration loss
• Optimization setting • Range and cost of optimization optimi zation parameter • KPI range and weight • Define AOI, UE, environment
• ACP (automatic (automatic cell planning) output output • Optimize site config config per sector level (e.g., height, azimuth power, power, type, m-tilt, e-tilt) • Compare network analysis before/after optimization 2
summary Use planet v5.3 • Aim • Convert NSN Hangzhou LTE TDD trial project • Optimize site based on some KPI example
• Input • • • •
2D maps (clutter, height, clutter height) 3D maps (building polygon with AGL) NSN project data (site long/lat, PCI, freq, antenna, NL) Default Default site config (power, (power, loading) loading)
• 3D Propagation model • UM model at 2.6GHz with 2D + 3D maps • With default building penetration loss
• Optimization setting • Range and cost of optimization optimi zation parameter • KPI range and weight • Define AOI, UE, environment
• ACP (automatic (automatic cell planning) output output • Optimize site config config per sector level (e.g., height, azimuth power, power, type, m-tilt, e-tilt) • Compare network analysis before/after optimization 2
Inpu In putt map map (1) (1) – ra rast ster er map maps s
• 5m, 20m resolu resolutio tion n maps maps • Height, Height, clutter clutter,, clutter clutter height height • converted converted from from BIN BIN file file format format to vertic vertical al mapinfo mapinfo mapp mapper er form format at (gr (grc c and and grd grd + TAB TAB)) • all maps maps have have same projec projection tion (Gauss(Gauss-Kruge Krugerr 117) 117) 3
Inpu In putt map map (2 (2)) – ve vect ctor ors s
• boundary, boundary, airport, street, subway, subway, Mstreet, Oroad, expressway, expressway, railways etc. • converted converted from from ASCII ASCII file file format format to vertical vertical mapinfo mapinfo mapper mapper format format (MAP (MAP, ID, DAT + TAB) • all maps have have same projecti projection on (Gauss-K (Gauss-Kruger ruger 117) 4
Input map (3) – 3D building map
• converted from ASCII file format to vertical mapinfo mapper format (MAP, ID, DAT + TAB) with mapped column for polygon_ID, AGL (i.e., float type) • all maps have same projection (Gauss-Kruger 117) 5
Input data (1) – site config
Convert NSN excel file into planet excel worksheet (site, antenna, sectors, sector_antennas etc)
6
Input data (2) – neighbor list
Convert NSN neighbor list excel file into planet format excel
7
Input antenna (1) – comba files
Convert comba antenna pattern and group mult-band, multi-etilt pattern into 1 planet antenna file format (.paf) • ODS-090R15NV06(F) • ODS-090R15ND06(F)
8
• max etilt = 6 • used freq range = 2.6GHz • x-polar with 4 antenna column (i.e., 8 port)
Input antenna (2) – conversion
• based on given etilt=0 and 6, interpolate etilt pattern in between (1..5] • assume +/-45 pattern is same for each etilt/band combination • use 65 deg broadcast azimuth BW (for now) 9
Converted planet project
• 2 group of sites (hangzhou and xiasha) • total 48 sites, 141 sectors • assume only 1x 5MHz 2.6GHz carrier per each sector 10
Spectrum and frame config – (1) • •
• •
11
EARFCN=37750+300=38050 or 2600MHz for LTE TDD BW = 5MHz
Use default set of MCS bearer 2 methods in planet (single value CINR or use spectrum efficiency curve), use later method in this case study
Spectrum and frame config – (2) Assume no ICIC
Assume • frame config has 12 DL slots “DSUDD_DSUDD” • S-subframe config “5”
DL overhead config (PDCCH) UL overhead config (DRS, SRS, PUCCH)
% DL RE used for overhead (CP + PDCCH/PCFICH/PCHICH + PBCH + RS + PSS + SSS) for different # tx antenna
12
Spectrum and frame config (3) For slow fading, assume spatial correlation between best serving sector and interfering sector is considered in CINR estimation
CINR Standard Deviation Scenario
Correlated slow fading
Noise limited areas
Standard deviation of signal strength
Interference fully correlated with server 0 dB
13
Areas between co-site sectors
0 dB
Interference not correlated with server
Based on correlation between signals
Default site config Loading % = 50 (DL), UL (20) UL noise rise = 1.5dB
PA power = 43dB (before any splitter) No power boost for RS, PSS/SSS
14
3D propagation model – (1)
• • •
15
3D model known as universal model (UM) from France Telecom – orange lab Freq = 2600MHz Rx height = 1.5m (can change for different building level for indoor coverage)
3D propagation model – (2) 5m, 20m, height
5m, 20m, clutter map with Clutter type and approx clutter height [for pixel where clutter height map is not available]
5m, 20m, Clutter height
2D raster map
16
3D propagation model – (3)
3D building map
17
3D propagation model – (4) Facet represents reflection from far away large obstacles
Morphologies represent mapping of clutter class to UM clutter class for customized optimization
UM specific generated data
18
3D propagation model – (5)
Graph represents horizontal guided propagation direction to account for horizontal diffraction either on side OR top of building 19
3D propagation model – (6)
Used model defined building penetration loss (outside/inside, inside/inside) , which depends on tx/rx path length, angle, freq <3km, UM prediction resolution = 5m (geodata map) >3km, UM prediction resolution = 5x2m=10m
20
3D prediction – RSSI example Hangzhou site example
3D prediction shows • attenuated indoor coverage due to penetration • scattering by building • Waveguide effect along narrow street 21
xiasha site example
Optimization setting – range Relative means with respect to current NSN setting Azimuth = relative [-30, +30] deg M downtilt = relative [-5, +5] deg Power = relative [-5, +5] dB Antenna height = fixed at NSN setting
Optimize e-tilt = absolute range [0..6] deg Optimize antenna pattern by picking either from antenna group “NSN_comba” which contains 2 antenna, ODS-090R15NV06(F) , ODS-090R15ND06(F) Optimization range is defined independently per sector 22
Pre-optimization (1) – define AOI
• AOI (area of interest) is defined for site group “hangzhou”. • Optimization in planet will be restricted to site group “hangzhou” within this AOI.
23
Pre-optimization (2) – define UE
Define UE equipment • no monte carlo simulation is performed in this case study • set usual UL parameter (e.g., 0dBi antenna, all bearer suported in UL, 24dBm)
24
Pre-optimization (3) – define environment
• each pixel is split into up to 4 environment • each environment has its related parameter such as speed and fast fading margin • some clutter has certain clutter disabled (e.g., no indoor/deep indoor for water related clutter) 25
Pre-optimization (4) – analysis
26
• run network analysis using NSN default site config (i.e., before ACP) • set RSRP threshold = -105dBm • use FULL UL power control (i.e., UL tx power is reduced until CINR for required MCS bearer is reached) • UE speed = 3km/hr • use linear PoC vs loading % curve to compute co-channel interference • target cell edge coverage prob = 85% for outdoor environment
Optimization (1) – define profile Combination of optimization KPI that can be selected
set profile as combination of different optimization goals, e.g., • (weight=2) RSRP coverage with balanced footprint • (weight=1) spectrum efficiency
27
Optimization (2) – define scenario
Define optimization scenario • choose optimization ONLY • choose optimized sector = considered sector
28
Optimization (3) – run scenario Set optimization to area = AOI, environment = indoor (wt=1) and outdoor (wt=2)
29
Sector loading used in optimization • if optimization KPI is dependent on loading, it requires traffic map from AOI area • static load (constant) OR dynamic load (fluctuate during ACP process)
ACP – setting summary Profile 1 KPI Profile 1 loading dependent
KPI
Profile 2 Profile 2 loading dependent
ACP result 1
RSRP>105dBm
no
NA
ACP result 2
RSRP>105dBm with balance footprint
no
Max Spectrum efficiency with bin weighting by traffic map
Check ACP result
Compare RSRP before/after ACP Yes, create a default traffic map
Compare balanced RSRP AND DL max data rate before/after ACP with same loading %
Do 2 ACP scenario • for both indoor environment (weight=1), outdoor environment (weight=2) • first scenario with load independent KPI, e.g., max area % for RSRP > 105dBm, and compare with NSN RSRP layer
30
• second scenario with a default traffic map and max area % for both RSRP > -105dBm and spectrum efficiency and compare both RSRP and DL max data rate with NSN case
ACP 1 – RSRP
• • • • • 31
Optimization done in step 0..20 ACP finished in 1 min report shows progressive RSRP gain % report shows site config changes apply optimized site config at final step
ACP 1 – RSRP comparison After ACP (outdoor environment)
32
before ACP (outdoor environment)
ACP 1 – RSRP statistics (outdoor)
Area % within AOI with RSRP > -105dBm
Before ACP
after ACP • ACP gives about 15% area gain with RSRP above KPI • area % outside range represent NULL pixel value (e.g., no best server available, other 3 environment – indoor, deep indoor, vehicular) 33
ACP 2 – create traffic map
Default traffic map • traffic map area = optimization AOI • assume total # subscriber = 500 within AOI • assume some clutter weight
34
ACP 2 – loading used
Use same static loading as per sector (i.e., 50% DL loading)
35
Assume • max 1000 sub per 1x carrier (i.e., within 500 max sub limit) • min RSRP required = -105dBm
ACP 2 – result
• Optimization done in step 0..22 • ACP finished in 15 min • report shows progressive RSRP gain % and spectrum efficiency • report shows site config changes • apply optimized site config at step 5
36
ACP 2 – RSRP After ACP (outdoor environment)
37
before ACP (outdoor environment)
ACP 2 – RSRP statistics
Area % within AOI with RSRP > -105dBm
Before ACP
after ACP ACP gives about 18% area gain with RSRP above KPI
38
ACP 2 – RSRP balance Best serving sector layer Use numeric grid filter to generate histogram of RSRP per different best serving sector
RSRP layer
after ACP
39
ACP config gives a higher and more balanced RSRP per each best serving sector area
Before ACP
ACP 2 – max DL data rate After ACP (outdoor environment)
40
before ACP (outdoor environment)
ACP 2 – max DL data rate statistics
Area % within AOI with DL max data rate at range 0, 1, 2, 5, 10Mbps Before ACP
after ACP ACP gives area gain different for different max DL data rate range • 0~1Mbps à area gain 3.7% • 1~2Mbps à area gain 2.1% • 2~5Mbps à area gain 3.5% • 5~10Mbps à area gain 0.6% 41
ACP 2 – DL max spectrum efficiency After ACP (outdoor environment)
42
before ACP (outdoor environment)
ACP 2 – max DL spectrum efficiency statistics
Before ACP
ACP gives different area % for different DL spectrum efficiency range • 0~1 à 6.3% gain • 1~2 à 2.7% gain • 2~5 à 0.6 % gain • 5~10 à same 43
After ACP
ACP 2 – DL CINR After ACP (outdoor environment)
44
before ACP (outdoor environment)
