Mentum_LTE_planet_v521_07sep2010_peter.pdf

Application note for LTE FDD for planet v5.2.1 (build 44)

Peter Cheung, Technical ConsultantMentum HK09 Sep 2010

2

Work flow• Project setup

• Map, preference (e.g., unit, display) and project setting• Network setting (spectrum, mod CINR, frame setup)• Create site table (e.g., ASP, Capesso, network overlay)

• Sector setup• Organize sectors (flag/group/query/site set/site template)• General (long/lat, antenna, link budget, model, filter mask) • Antenna algorithm (MIMO, diversity, AMS)• Power (EIRP, avg per RE per channel)• ICIC related (RSRQ threshold and outer cell RE %)• PCID and A3 HO threshold

• Network analysis• Setup sub (equipment and service) and environment• Define area and PoC curve• other system interference (e.g., other system interference, Femtocell)• Generate layers for different loading %• Do statistics report, compare with KPI

• Monte Carlo• Generate traffic map and setup sub type• Setup scheduler and run MC for combination of sub type/environment• Apply cell loading and re-run network analysis

• Others• Generate IM, NL (single/multi-tech)• AFP and PCID

3

Project setup (1) – map DTM/cutter/clutter height/polygon/vector must have same projection/resolution (eg UTM xxx at xxx m)

Since model is map dependent, so need to select proper map to run prediction for different models

Map should be in UTM zone xxx and site/sector/DT should be in long/lat

Subfolder names under Geodata must be matched

4

Project setup (2) – preference

5

Project setup (3) – network setting Define spectrum (start/end DL/UL freq and BW)

For subband (e.g., with carriers gap), need to define separate band, since each sector can ssign ONLY one band

“grey” out means some sectors are using defined band. If need to modify existing band, need to first switch all sectors using this band to a “dummy” band first.

6

Project setup (4) – network setting Option 1 à CINR threshold for each MCS with one spectral eff[bits/symbol] at some specified PER %, bits/symbol can be extracted from CQI table in 3gpp 36.213

Speed of each MCS is defined in project setting

Option 2 à define spectral efficiency curve for each MCS. If used, new layers can be generated (e.g., UL/DL max/avg spectral eff) and affect max UL/DL data rate layers, best available DL/UL mod and UL noise rise (select max spectral eff)

7

Project setup (5) – network setting If Interference coordinate (i.e., ICIC) is enabled for this frame config, FFR, RSRQ threshold and RE % can be specified

carrier BW will defineFFT size, sampling freq

• # occupied DL/UL subcarriers (excl DC/guard)• # total RB/slot• # RB for sounding ref signal

Throughput (including L1 overhead)= #occupied subcarrier x #symbol/slot x #slot/frame x fps x bit/REe.g., 10MHz carrier with spectral eff=5bits/RE has 601x7x20x100x5 = 42Mbps

Throughput (excluding RS symbol)= #data symbol/RB x #slot/frame x fps x #RB/slot x #bit/symbole.g., 10MHz carrier with 4 RS symbol and 64QAM/no FER has 80 x 20 x 100 x 50 x 6 = 48Mbps

8

Project setup (6) – network setting DL frame editor UL frame editor

DL overhead % = total overhead / DL frame durationWhereTotal overhead = CP + PDCCH + PBCH + ref signal

Note that #RE allocated for PBCH and ref signal is dependent on # of tx antenna.

UL overhead = demod ref signal + PUCCH + sounding ref signal

9

Project setup (7) – use ASP

• Define basic (fit hexagon) or advanced (can use candidate site list, traffic map, coverage/model criteria)• Define clutter exclusion, antenna height range, max pathloss, site radius• Define polygon/area

Propagation model criterion

Traffic/site template criterion

10

Project setup (8) – use ASPOptimization constraint used by ASP and ACP (e.g., capesso) as input setting

Range of antenna height that can be adjusted by ASP

11

Project setup (9) – use ASP

Sites generated from advanced ASP (meet criteria for coverage then traffic, use customized site template for different clutter and with antenna height adjusted)

Sites generated from basic ASP (fit # polygon of defined radius for 1 site template for all clutter and with clutter exclusion)

12

Project setup (10) – use capesso

Create site candidate lists in planet first, run prediction. Then export to capesso and let capesso pick # site according to user defined criterion (coverage/capacity)

13

Project setup (11) – network overlay

• pick data source for other technology (e.g., group of GSM sites from same project)• select LTE FDD band used for newly created co-located LTE FDD sites (with same antenna ht/azimuth, # of sector etc)• save new LTE FDD site tabe and/or add suffix to sector ID [optional]

14

Sector setup (1) – organize sectors

Flag, group, query to organize sectors for display/network analysis

Site set is storage of full/partial set of different version (e.g., site set before and after optimization)

Query offer flexible sector group accordingly to user defined criteria

Site template can be created from existing sites

15

Sector setup (2) – generalSite à base station (different technology) àsector à antenna ID (e.g., split sector)

Each sector assign one band with its flag/group

Max pooled throughput of site (blocked sub in MC runs)

16

Sector setup (3) – general

Each sector assign• antenna algorithm• prediction model/distance/# of radial• one or multiple antenna with specified link budget and feeder length• Tx/Rx filter mask (at eNodeB)

Set threshold for blocked sub in MC run (coverage, #sub, UL noise rise)

Affect adjacent channel interference

17

Sector setup (4) – antenna algo

Smart antenna/beamformingàincreasing CINR by increasing C and decreasing I

Diversity gain à adding CINR margin

MIMO or spatial MUX à increasing throughput relative to SISO with option for UL collaborative MIMO and LOS exclusion [MIMO improve throughout, not necessary coverage]

Automatic MIMO switch à switch to diversity (low CINR) or MIMO (high CINR) or MIMO+diversity (high CINR with spare antenna for diversity)

See antenna algorithm application note for details

18

Sector setup (5) – powerCombined PA power (e.g., if #Tx antenna=2 with 40dBm each, PA power=43dBm)

Enable hopping can improve reference CINR, but also affect DL loading % to other cell depending on RE reuse

Ref signal power = PA power x (% of ref signal within 1 RB)Where % depends on #subcarrier for ref signal and dataE.g., 10MHz carrier has #ref signal subcarrier/RB=2, #total subcarrier/RB=12 à ref signal Tx power = 34.55dBm with 3dB boost

Avg power per RE for xxx signal (used to calculate different RSRP, DL CINR)

Affect UL CINR

19

Sector setup (5) – ICIC related• Input for network analysis, output for MC • DL loading % = traffic loading = used # data RE / available # data RE

e.g., 10MHz has 2 ref sign RE/RB, if 4 data RE/RB is used,then DL loading = 4/(12-2)=40%

UL noise rise = 10*log[1/(1-UL loading %)]

FFR usage % = % traffic loading in outer cell (only for frame config supports ICIC)

Multiple antenna % = % traffic support AAS

RSRQ (ref signal received quality) used to define inner/outer cell boundary

Outer cell % = RE% used in outer celle.g., 100% means no ICICe.g., <100% means PoC is used for co-channel interference

20

Sector setup (6) – PCID and HOPCID = [0..503] used for sector ID, and is combination of • phy layer cell ID group =[0..167] from SSS• phy layer identity = [0..2] from PSC

Since ref signal hopping is dependent on PCID, if PCID of best server/interferer is same, ref signal CINR is lower

Used for AFP

Threshold to trigger A3 measurement if neighbor sector has xxdB higher than serving sector

21

Network analysis (1) – setup sub

UE/CPE Equipment (e.g., mobile, USB dongle)

Hardware à UL power parameters

UE/CPE antenna à pattern, MIMO, polarization diversity (i.e., x2 Tx port)

Supported UL/DL modulation bearer

22

Network analysis (2) – setup subPacket switched service (e.g., VoIP, 64k email, 1M web/FTP, 128k email) with different priority and QoS

Convert traffic map to #sub to spread by MC by individual loading specified [see MC ppt for detail]

Actual throughput based on UL/DL activity %

QoS class priority (e.g., 1 à min=max data rate, 9 means min=0.1kbps (BE))

Sub type ONLY use in MC (see later slides on MC)

23

Network analysis (3) – environment

Planet has 4 default environment (I, O, V, DI)• need to map clutter class to 1~4 environment• e.g., water related clutter has no I, V, DI, then that analysis on non-mapped environment will not consider that clutter class• fast fading margin = headroom for Rayleigh fading variation (e.g., 60km/hr require 10dB for 64QAM)• penetration loss = extra building/vehicle loss apart from one from prediction model (e.g., 10~30dB depending on wall material)\• vehicular speed (V environment) = typical speed of UE/CPE for associated clutter

Clutter-dependent DL orthogonality %

24

Network analysis (4) – define area

Define area (e.g., rectangle, polygon)

Area grid can be used in network analysis/MC/analysis statistics report

25

Network analysis (5) – setting

Each network analysis has one• band• equipment• environment• cell edge probability % KPI, used in Jakes’s curve from slow fading margin(for MC, different % for different service)• speed• # UL RB per user (assuming each map pixel has 1 user)

Choose best server if• total received signal strength > threshold (i.e., sensitivity) • has highest ref signal strength [dBm]OR highest RSRQ [dB]

26

Network analysis (6) – ICIC input

ICIC uses to reduce reused RB in outer cell (i.e., improve cell edge coverage prob %)

• Basic = remove % interference based on DL loading % at best serving sector (i.e., without interaction between eNodeB)

• Advanced = sort interfering sector according to received power and rank sectors to prioritize interference removed by scheduler(i.e., most interfering sector will use non-interfering RB the most, with interaction between eNodeB)

See separate ICIC ppt for details

Under ideal channel conditions, OFDMA guarantee orthogonality between RB, so no co-channel interference for inner cells from best server

27

Network analysis (7) – UL PC

No DL power controlUL power control• FULL = decrease Tx of UE/CPE just enough to achieve the required CINR of given MCS

• fractional = further decrease Tx of UE/CPE by pathloss weighted by alpha= this prevent UE/CPE with large pathloss (i.e., along inter-cell boundary) to

transmit excessive power and interfere with all neighbor cells

3GPP equation for fractional UL power control

28

Network analysis (8) – define PoC

PoC % of reusing the same RB as a function of DL loading %. This PoC curve is used to calculate co-channel interference

e.g., increasing slope of PoC with scheduler

Normal PoC curve

29

Network analysis (9) – other system interference

Option 1 à as fixed power [dBm] in same channel per sector

Power setting for each sector

Option 2 à import interference grid file (.grd) with specified center freq/BW

E.g., CDMA1xRTT overlay • DL interference à import best server pilot Ec (since always Tx)• UL interference à UE EIRP

30

Network analysis (10) – Femtocell

Run Femtocell extension [see separated Femtocell ppt for details]

31

Network analysis (11) – layers

Layers can be divided to • Common (for all carrier)• per carrier

Best server related(total/ref signal/syn signal)

HO related

DL coverage (prob%, AAS, CINR, data rate, CQI, MCS, margin)

UL coverage (prob%, CINR, MCS, margin, data rate, tx power)

Summary coverage

32

Network analysis (11) – diff loading

DL CINR

75% DL loading25% DL loading

Similar to cell loading in CDMA network

Examine layer for different setting (e.g., DL loading %, antennalgorithm) during initial planning scenario

DL avg data rate

33

Network analysis (12) – ref signal• 3 analysis layers generated related to reference signal

• RSRP (ref signal received power) [dBm]= EPRE (energy per RE) in 3GPP= avg power received on 1 ref signal RE

• Ref signal strength [dBm]= total power received on ref signal RE= RSRP x 10log(# of ref signal RE)

• RSRQ (ref signal received quality) [dB]= 10log (N x RSRP / RSSI)

Where N = total #RB, RSSI=avg power received on ref signal symbol

• Max RSRQ = -3dB• UE/CPE measure RSRP and RSRQ• Compare to WCDMA

• RSRP à CPICH RSCP for coverage KPI• RSRQ à CPICH Ec/No for interference KPI

• Cell selection procedure [idle mode]à RSRP is used • HO procedure à can be RSRP, RSRQ or both

34

Network analysis (13) – statistics

Can do statistics on one or multiple layer

Since each layer is a mapinfo grid file, statistics report can be generated and check for KPI

Include NULL and/or zero values in statistics

35

Network analysis (14) – statistics

Can filter by one filter

Can filter by >1 filter

Similar to grid query from grid manager, extract statistics based on different filtering criteria

36

Network analysis (15) – statistics

Step size used for statistics

Step size used for statistics

% sub area and % sub within area X

Filter by grc grid (e.g., best server or clutter map)

Filter by traffic map (e.g., Er, #sub, throughput)

37

Network analysis (16) – statistics

Best server laye filter by numeric grid (e.g., DL avg data rate)

Compute mean/min/max/median/std dev/RMS of all bins within best serving coverage area of sector xx

Note, mean grid is NOT avg sector thoughput, since mean is avg is ALL DL data rate bin, assuming there is 1 sub per bin (e.g., problem for overshoot area with no sub)

38

Network analysis (17) –optimization based on statistics

[see separate ppt for LTE optimization based on statistics]

• Sort (ascending/descending)(e.g., find top 10 sectors with highest coverage prob %)• filter by specified range• generate statistics CDF/PDF• generate labels• generate sector display scheme (e.g., different sector color to rep. specified data rate range)

39

• Coverage for RS• Boost for RS• PCID and RS FH

• Power affects coverage/interference• DL power à RS and PSS/SSS signal with boost• UL power à PUCCH, PUSCH and SRS

• Antenna config• Multiple antenna affects DL CINR, which determine highest MCS, max data rate

and spectral efficiency• E-tilt/azimuth/height/type improve DL CINR for sub close to eNodeB

• ICIC• Affect DL CINR and data rate, especially at outer cell • Affect cell edge coverage probability, especially for high MCS• Static ICIC using 1:3 reuse• Dynamic ICIC based on channel/traffic conditions (e.g., RSRQ, FFR %, outer cell

RE %)• Scheduler

• Assign RB according to predefined rules [see MC slides later]

Network analysis (18) –optimization parameter examples

40

MC (1) – generate traffic mapGenerated Traffic map can be convert (between different unit), scaled and combined

Generate traffic map by• regions (polygon)• vector• classified grid (e.g., clutter map)• network data (e.g., recorded OSS traffic statistics per sector)

41

MC (2) – use traffic map as demand

Traffic demand for LTE can be #sub or throughput [kbps]

Tabulate site ID/sector ID/ # of sub as an excel file

Import excel file to planet and map column Import excel to operational data

NOTE: use MC for indication to capacity/dimensioning and network stability (i.e., when network becomes saturated and start to block sub)

42

MC (3) – generate TM via best server with clutter weight

Create best server grid based on RS

Apply weight for different clutter class

Spread traffic demand per sector according to best serving coverage bins

Combine vector to clutter as “new” clutter

43

MC (4) – generate TM via best server with clutter weight

Combine vector “primary” to original clutter as new clutter map “primary_clutter”, and a new clutter class “highways” is added

Generated traffic map

44

MC (5) – setup sub type

1 sub type have

• traffic map• equipment• one or more usage (weight for 4 environment) for one or more services/speed combination

Priority set for different sub type

45

MC (3) – setup MAC schedulerProportional demandProportional fairMaximum capacity

Scheduler types (serve sub at min data rate)• priority

•Based on priority in sub setting• proportional demand

•RB given to low data rate sub • max capacity

•RB given to high data rate sub• proportional fair

•RB given to sub with better CINR• user defined

•RB given to sub with user defined weight

46

MC (4) – settingChoose one or more sub type in 1 freq band

Input # of MC run and required convergence %

Choose scheduler

Steps of MC• spread sub randomly and independently per run• sort sub by priority (sub type, service, QoS)• analyze DL and UL, check if sub is served• generating operating point and sub info (e.g., status and spreading)

sub blocked # avg

minmax% 5 #run 5last of # runlastofsubblockedsubblockedeconvergenc

−=

47

MC (5) – generate reportStatus with blocked reason

Detailed MC report for sub/throughput per sector/carrier

48

Other (1) – generate IMUse IM as input to generate NL, AFP and PCID

Histogram based IM can be in area % or traffic % between server/interferer

49

Other (2) – import IM

50

Other (3) – generate NLNL can be generated • based on best server grid or IM• for single technology (LTE à LTE)• for multiple technology (LTE à CDMA, CDMA à LTE)

From Best server grid, get NL from border

From IM, get NL from overlapped area/traffic or both

51

Other (4) – edit/compare NL

NL result can be view in map or manually add/delete

Different NL table can be compared and merged

52

Other (5) – generate AFP and PCIDViolation cost for AFP

Violation cost for PCID

For AFP/PCID planning, IM is mandatory input, and NL input is optional

53

Other (6) – AFP and PCID results

Display PCID/carrier in map

Report of reuse distance/interference % for different sector

Backup slides

55

Spectral efficiency vs CINR

Different curve for different MCS under different channel conditions

56

DL frame structure

57

DL reference signal (3GPP 36.211)

Normal CP, # Tx antenna = 1

Normal CP, # Tx antenna = 2

Normal CP, # Tx antenna = 4

Depending on # Tx antenna, # and RE position is different for reference signal of different antenna

58

A3 Handover event

59

LTE QCI (QoS class ID)3GPP 23.203 QCI table for different services and PER