Lte Capacity Management v0.1 MBNL Esl

LTE CAPACITY MANAGEMENT
KPIs and Dimensioning rules
11/09/2014 Remus Sepp

v0.1 EE Confidential document
Agenda
1. Introduction and Scope
2. LTE Capacity Domains
3. LTE Radio I/F Capacity
4. LTE Transport Capacity
4a. Backhaul
4a1. eNB side
4a2. Network side
5. Annexes
EE TEMPLATE FOOTER 11/09/2014 2

1. Introduction and scope

1. Scope
This document provides capacity management guidelines for the life cycle of
LTE RAN system, including:
1. initial dimensioning of new cells design;
2. identification of overloaded and under-loaded cells and/or architecture
elements;
3. Re-dimensioning of capacity to more closely match current traffic and
seasonal traffic trends;
4. Interfaces dimensioning to fit the above mentioned LTE guidelines are also
included;
These guidelines are aimed at cell planners. Recommendations for parameter settings and use of
features such as traffic steering are not given as these are covered by RAN Performance group guides
and established practice in the regional groups.
Purpose
To establish a set of best practice guidelines for cellular capacity management
to be used in all Planning regions in order to ensure a consistent approach to
cellular
EE dimensioning.
TEMPLATE FOOTER 11/09/2014 4
2. The Capacity Management cycle
Identify
overloaded
elements
Monitor
Analyse
utilisation
trends
Implement Select
changes solution

2. The Capacity Management cycle (continued)
The Capacity Management Loop 3. Trend analyses: this step is based on the regular
LTE forecast which will offer traffic guideline
follows the usual pattern used in other figures that can be envisaged in future; the BAU
Radio Access Technologies, however dimensioning results should be future proof and
with specific measures and limitations ensure enough levels of capacity are being
characteristic to LTE: delivered for at last ½ years in advance;
1. Monitor Utilisation: by using the specific 4. Solution: in the current network-maturing stage
measures recommended in this document and the options to expand the RAN elements are quite
the tools available to the EE Planner it is easily limited – however, this will change as the network
possible to build a comprehensive picture for gets more and more mature, so this step will
the entire network’s capacity; become critical for the adjacent costs;
2. Identify the overloaded elements: LTE is a 5. Implementation: this step is included in the
complex system, so it is to be expected there previous picture for completeness, however the
are several elements/nodes that can get current document doesn’t include any details in
congested: PRBs, PUCH/PRACH/PDCCH this respect, as this is more of a System Design
resources, SRS and paging, LMPTs/LBBPs and Planning activity.
CPUs, S1 or X2 User planes, Ethernet, etc.
This document treats each of these potential
elements and proposes rules/thresholds that
will prevent the system to get in congestion;
3. Traffic measures and definitions
Scope
To define a precise set of measures that will be used in the LTE Capacity Management process.
0. SLAT: Service Level Attention Threshold  threshold value used in the LTE Capacity Management
process to trigger the corresponding action on the node element.
1. Daily Busy Hour traffic:
This is a per-cell measure used to characterize the carried traffic in DL or UL; is defined as the average
traffic throughput carried per cell during the time when the cell’s traffic was maximum in a day.
Measurement unit: [Mbps];
Example: 61157B11 has for 21/June/2013 a Daily BH traffic of 16.07 [Mbps] in DL and 11.03 [Mbps] in UL.
2. Top Daily Value (for ratio KPIs):
This is a per-cell or per-site measure used in the LTE Capacity Report and is actually linked to each
performance KPI that is expressed as a ratio and is used in the LTE Capacity Measurement process; it
does express the de-sensitised(1) maximum daily value for that respective KPI, where the “de-sensitised”
notion means that the Top Daily Value is calculated for all the daily samples which have significant high
denominator.
Measurement unit: [%];
Example: On the 23/July/2013, 98530C11 has a “Top Daily Value” of E-RAB Resource Congestion Rate of
1%  calculated as “(E-RAB Failed Establishments due to no Radio resources) / (E-RAB Attempts)” and
where E-RAB Attempts>50
(1) the “de-sensitising” condition is put in place to eliminate ratio KPI values that are calculated based on very small values in the denominator . 7
EE TEMPLATE FOOTER 11/09/2014
3. Traffic measures and definitions (continued)
3. Daily Busy Hour KPI value:
This is a per-cell/per-site KPI measure used to characterize the KPI in discussion for the daily period when
the traffic was in BH;
Measurement unit: [the same as the constituent KPI];
Example: 61157B11 has for 10/June/2013 a Daily BH RRC_Congestion_Rate of 0.3%.
4. Top 3 Weekly Value (T3W value):
This is a per-cell or a per-site KPI value meant to express a “consolidated” statistical view for an entire
week. The meaning of “consolidated” is that this KPI’s trend and evolution across medium and long term
scale will bring significant added value into the Capacity Management process.
The calculation method for this type of Traffic measure: it is the average of the Top 3 daily values of the
KPI in discussion calculated for a single week, for a single cell or a single site.
This is a weekly measure, so it is recommended to report it on each Monday, such as the calculated value
is reflecting the whole week in discussion.
Measurement unit: [the same as the underlying KPI];
Example: For the week commencing 05/August/2013, 98530C11 has a “Top 3 Weekly Value” of E-RAB
Resource Congestion Rate of 1.37%  this is the result under the assumption that the cell in discussion
has an average of 1.37% for the highest 3 days in terms of top daily values of E-RAB Resource Congestion.

2. LTE Capacity Domains

For a systematic approach, the following LTE Capacity
domains have been identified:
LTE capacity domains I. LTE Radio (eNodeB and Radio interface);
II. LTE Transport domain which can be split into:
a. Backhaul domain (between eNB and MSN PE)
b. Connectivity domain (between the MSN and the
core PE routers)
Web III. Infrastructure domain (PKI and DHCP servers)

SGW
eNB
CPCN/
GRN/
core PE MSN PE
MPLS Switch sites MSN
MME network network
IIb. Connectivity
domain IIa. Backhaul domain
eNB
PKI
II. LTE Transport I. Radio
DHCP I/F
User data domain
III. Infrastructure
Signalling data
domain
LTE capacity domains – tree architecture
LTE Capacity
Domain
Radio I/F Transport
Site Cell Backhaul Connectivity Infrastructure
Sec-
eNB side Network side AR DHCP PKI
GW

3. LTE Radio I/F capacity

a. General Radio I/F aspects

LTE Radio I/F - Congestion points
The congestion points can be split in two parts: cell and site congestion points:
c2. DL/UL PRB Utilisation
c6. Paging
Resources c7. Flow control
eNodeB
c5. PDCCH Resources
MRFU
Transport LMPT LBBP

CPU CPU
c4. PUCCH Resource
Usage s1. Connected
LTE users
c3. SRS Resource

s2. (L/U)MPT Usage c1. PRACH
CPU/Mem Resource
usage Usage
s3. LBBP CPU/Mem. Radio interface domain
usage
LTE Radio: Channel mapping

LTE Radio I/F – Site cabinet/board info
1. FIND THE INSTALLED BOARDS
MML
2000
command
2. FIND THE TYPE OF BOARD ONCE SLOT NUMBER AND SUBRACK NO. KNOWN
MML
2000
EE TEMPLATE FOOTER command 11/09/2014 16
Baseband boards – LBBP:
LTE Radio: Typical capacity figures: Limit LBBP-c LBBP-d1 LBBP-d2
Limit UMPT LMPT
Main control boards: #Cells 3 3 3
Max 1500 DL / 450 DL /
In-sync RRC 400/cell, 400/cell, 400/cell,
Mbps/module 1500 UL 300 UL
users 600/card 1200/card 1200/card
In-sync RRC Out-of-sync 800/cell, 800/cell, 800/cell,

3600 1800
users RRC users 1200/card 2400/card 2400/card
Out-of-sync DL/UL max 225/100 450/225

7200 3600 600/225
RRC users throughput Mbps Mbps
MIMO 2x2 2x2 4x4, 4rxdiv

LBBPc card with RRC Capacity Limits per elements combined:
Out-of sync RRC connected
Number Number In-sync RRC connected users
Number users
of of cells
of LTE Max total Max total
LBBPc at Max in Max in Max per Max in
bands across all across all
cards 2x20MHz one cell 1xLMPT cell 1xLMPT
LBBPc LBBPc
1xLBBPc 3 1 600 1200
2xLBBPc 6 2 400 1800 800 3600
1200 2400
3xLBBPc 9 3
eNB Cards config.:
LBBPc Card with Maximum Throughput Capacity
Max Downlink Throughput
Number Max Uplink Throughput (Mbps)
(Mbps)
Number of cells Number
of LBBPc at of LTE Max per Max per
Max total Max total
cell Max in cell Max in
cards 2x20MH bands across all across all
(10/20MH 1xLMPT (10/20MH 1xLMPT
z LBBPc LBBPc
z) z)
1xLBBPc 3 1 300 100
2xLBBPc 6 2 73/150 750 43/70 750

450 200
3xLBBPc 9 3
b. Radio site aspects

LTE Radio I/F – site congestion points
Impact of lack of Monitored
No: Resource type Meaning Limit Solution
resources item
Add licenses if
No new users, Connected (L/U)MPT CPU
Connected users Max users in
degraded Users Usage <60%, add
s1 per (L/U)MPT card RRC_Connected >60%(1) eNodeBs if
experience for License
(1 per site) mode (L/U)MPT CPU
existing users Usage
Usage >60%
Processing Expand the

(Avg CPU Usage) >60%
Main control board capability of the Bad KPIs, bad LMPT/UMPT eNodeB
s2 or (%time “CPU Usage
(LMPT/UMPT) CPU main control experience CPU load Control plane
>85%” is >5%)
board capacity
LTE baseband
Processing Bad KPIs, bad (Avg CPU Usage) >60% Expand the
processing unit LBBP CPU
s3 capability of the experience or (%time “CPU Usage eNodeB User
(LBBPc/LBBPd1/ load >85%” is >5%)
main LBBP board plane capacity
LBBPd2)
(1)  this is a “per-eNB” figure

PRS
LTE Radio I/F – site congestion KPIs
Meas. Aggreg
KPI Index KPI name Formula Unit . SLAT
Object
LBBP-c 300
Average LBBP in-sync Σ(L.Traffic.User. Ulsync.Avg), across all cells
R.S.U.H1 #users T3W (1) LBBP-d1 300
Connected Users belonging to one LBBP card
LBBP LBBP-d2 300
card LBBP-c 480
Σ(L.Traffic.User.Avg - L.Traffic.User.
Average LBBP out-of-
R.S.U.H2 Ulsync.Avg), across all cells belonging to #users T3W (1) LBBP-d1 480
sync Connected Users
LBBP card LBBP-d2 480
Upgrade path:
R.S.U.H1,2 violated 3/5 weeks  follow the LBBP expansion guidelines
(described further below);
(1)  T3W = Top 3 Weekly value,
defined in the Introduction
(2)  see the LBBP users limitations
table at the beginning of this
Primary KPI chapter
Secondary KPI
11/09/2014
EE TEMPLATE FOOTER 20
PRS
Meas. KPI
Object INDEX
KPI name Formula Meas. unit
Aggreg. SLAT
LMPT/UMPT module Σ(L.Traffic.User.Max), per (L/U)MPT Weekly LMPT 1080

R.S.U.H3 #user
connected users module Max UMPT 2160
Σ(L.Thrp.bits.DL)/10^6/3600 per cell, LMPT 270
LMPT/UMPT module total DL Weekly
R.S.U.H4 across all cells in the eNB (L/M)MPT [Mbps]
throughput Max UMPT 900
module
Σ(L.Thrp.bits.UL)/10^6/3600 per cell, LMPT 180
LMPT/UMPT module total UL Weekly
R.S.U.H5 across all cells in the eNB (L/M)MPT [Mbps]
throughput Max UMPT 900
module
R.S.C.H1 LMPT/UMPT CPU load VS.Board.CPUload.Max (Slot No. 7 only) [%] T3W 50%
Site
Percentage of times that the
LMPT CPU usage reaches or VS.Board.CPULoad.CumulativeHighl Weekly
R.S.C.H2
exceeds a preconfigured oadCount/3600 x 100%, slot 7 only
[%]
Max
85%
threshold
LBBP CPU load (on each
VS.Board.CPUload.Max (Slots No.
R.S.C.H3 individual LBBP card separate
0…4 only)
[%] T3W 40%
CPU load)
Percentage of times that the
LBBP CPU usage reaches or VS.Board.CPULoad.CumulativeHighl Weekly
R.S.C.H4
oadCount/3600 x 100%, slot 7 only
[%]
Max.
85%
exceeds a preconfigured
threshold
LMPT – slot 7;
Please refer to Annex 2 LBBP – slots 0…4 (1) – the “DL per cell throughput” KPI is defined in the
EE TEMPLATE FOOTER 11/09/2014 cell congestion KPI tables further below 21
PRS
• Huawei is recommending to keep en eye on the Call Attempts per Second: CAPS
• The CAPS formulas is a composite formula, having 8 components as per the table below:
CAPS = w1 x “PS_Calls_with_S1_release” + w2 x “Paging” + w3 x “TAU” + w4 x “Intra_eNB_HO” + w5
x “Inter_eNB_HO” + w6 x “Inter-RAT HO” + w7 x “CSFB” + w8 x “RRC Connection Fail”
• There is a separate set of weighting factors (w1…w8) per each board type: LBBP-c, LBBP-d and LMPT;
Traffic Mode Counter of the Report_Period LMPT Equivalent weight LBBPc Equivalent weight LBBPd Equivalent weight
PS Call with S1
L.E-RAB.AttEst 1 1 1
release
Paging L.Paging.S1.Rx 0.0221 0.0373 0.0403
TAU L.RRC.ConnReq.Att.MoSig 0.5196 0.6262 0.6269
L.HHO.IntraeNB.IntraFreq.PrepAtt
Out" +
Intra_eNodeB HO 0.5456 1.0602 1.139
"L.HHO.IntraeNB.InterFreq.PrepA
ttOut
L.HHO.IntereNB.IntraFreq.PrepAtt
Out" +
Inter_eNodeB HO 1.5973 1.1645 1.2526
"L.HHO.IntereNB.InterFreq.PrepA
ttOut
("L.IRATHO.E2C.PrepAttOut" +
"L.IRATHO.E2W.PrepAttOut" +
Inter-RAT HO 0.8446 0.4138 0.4
"L.IRATHO.E2G.PrepAttOut" +
L.IRATHO.E2T.PrepAttOut")
CSFB L.CSFB.PrepAtt 1.5471 1.4757 1.5088
("L.RRC.ConnReq.Att" -
RRC Connection Fail 0.2635 0.2672 0.2444
"L.RRC.ConnReq.Succ" )

PRS
• Huawei CAPS limits per board:
Software Version
CAPS supported (call eRAN7.0/SRAN9.
attempts per second) eRAN6.0/SRAN8.0
eRAN3.0/SRAN7.0 0
LMPT 30CAPs 30CAPs 60CAPs

UMPTa6 60 CAPs 120 CAPs 120 CAPs
UMPTb 60CAPs 120 CAPs 150 CAPs
Hardware
LBBPc 30CAPs 30CAPs 60CAPs
Board
LBBPd1 60CAPs 60CAPs 80CAPs

PRS
LTE Radio I/F – site congestion KPIs - CAPS
Meas.
KPI Index KPI name Formula Unit Aggreg. SLAT
Object
Average LBBP-c Σ(CAPS_Formula)/3600, across all cells Call atts

R.S.C.H5 per sec.
T3W(1) LBBP-c 20
1 LBBP CAPS belonging to one LBBP-c card
card Average LBBP-d Σ(CAPS_Formula)/3600, across all cells Call atts
R.S.C.H6 per sec.
T3W(1) LBBP-d 40
CAPS belonging to LBBP-d card
1
Average (L/U)MPT Σ(CAPS_Formula)/3600, across all cells Call atts
(L/U)MPT R.S.C.H7 per sec.
T3W(1) (L/U)MPT 80
CAPS belonging to (L/M)MPT card
card
Upgrade path:
R.S.C.H5,6 violated 3/5 weeks  follow the LBBP expansion guidelines (described
further below);
R.S.C.H7 violated 3/5 weeks  to be defined
Primary KPI (1)  T3W = Top 3 Weekly value,

defined in the Introduction
Secondary KPI table at the beginning of this
chapter
11/09/2014
LTE Radio I/F – site congestion expansion steps
Meas. KPI Meas.
KPI name Violation Expansion step
Object Index unit Aggreg. SLAT
LMPT/UMPT module LMPT 1080

R.S.U.H3 #user Weekly Max 5/9 weeks
connected users UMPT 2160
LMPT/UMPT module total LMPT 270

R.S.U.H4 [Mbps] Weekly Max 5/9 weeks
DL throughput UMPT 900
Add an eNodeB and
LMPT/UMPT module total LMPT 180 connect it to the
R.S.U.H5 [Mbps] Weekly Max 5/9 weeks evolved packet core
UL throughput UMPT 900 (EPC) through a new
S1 interface
R.S.C.H1 LMPT/UMPT CPU load [%] T3W 60%
Percentage of times that

Site 3/5 weeks
the LMPT CPU usage
R.S.C.H2 reaches or exceeds a [%] Weekly Max. 85%
preconfigured
threshold
LBBP CPU load (on each

R.S.C.H3 individual LBBP card [%] T3W 60%
Follow the LBBP
separate CPU load)
expansion process
Percentage of times that 3/5 weeks presented further
the LBBP CPU usage below
R.S.C.H4 reaches or exceeds a [%] Weekly Max. 85%
preconfigured
threshold
LTE Radio I/F – LBBP expansion steps

LTE Radio I/F – eNB Capacity KPIs: example stats

c. Radio cell aspects

LTE Radio I/F – cell resource allocation
• When a UE requests RRC connection establishment for a
service admission or a handover, the eNodeB allocates
CQI resources, SRI resources, and semi-persistent ACK
resources (if in semi-persistent scheduling mode) for the
UE. If the allocation of SRI and CQI resources fails, the
UE admission fails;
• If a UL unsynchronized UE requests uplink

synchronization for data transmission, the eNodeB
allocates CQI resources, SRI resources, and semi-
persistent ACK resources for the UE. If the allocation of
SRI and CQI resources fails, the UE cannot restore uplink
synchronization and therefore fails to transmit data;

LTE Radio I/F – cell congestion points - Huawei solutions
Impact of lack
No: Resource type Meaning Monitored item Limit Solution
of resources
Enable adaptive backoff
Usage of or resource adjustment
contention-based algorithm for PRACH,
pre-ambles > 75% otherwise split the
carrier/sector
PRACH Delayed/failed PRACH resource
c1 UL initial access
resources(1) initial access usage Enable the resource
adjustment algorithm
Usage of non-
for PRACH and reuse of
contention-based
dedicated pre-ambles
pre-ambles > 75%
between UEs, otherwise
split the carrier/sector
Failed >70% and

Bandwidth on Air I/F, Add carriers or
c2 PRBs admissions, bad DL PRB Usage User_DL_throug
DL only hput <2MBps eNodeBs
experience
Loss of radio
>60% and
capacity due to SRS resource Change the SRS
c3 SRS resources UL Scheduling info SRS algorithm
inaccurate DL Usage configuration
switch is OFF
measures
Failed >60% and Set PucchSwitch:ON

and reduce
PUCCH UL control channel accesses/HOs, PUCCH resource PUCCH
c4 UeRelDelayTimer, then
resources resources decreased UEs in usage algorithm split the cell, add more
the system switch OFF carriers or eNodeBs
(1) – please check the next slide for the PRACH

LTE Radio I/F – cell congestion points - Huawei solutions
Impact of lack
No: Resource type Meaning Monitored item Limit Solution
of resources
Set PDCCH Symbol

DL control channel Bad user PDCCH resource CCE Usage > Number Adjust
c5 PDCCH resources Switch:ON if DL PRB
resources experience usage 80%
Usage <90%,
%Paging
received on S1 Decrease #cells in
Delayed
Paging eNodeB paging Paging > 60% of the Tracking Area the
c6 sessions, bad congested cell
resources capacity resource usage paging
experience belongs to
messages
>1500
User plane and

eNB controls the in/out This is a As per Huawei
Control plane
c7 Flow control flow to prevent preventive thresholds/par Expand the eNB
resources per
overload feature ameters
eNB

LTE Radio I/F – Huawei users/cell counters detail
• Because the eNB boards are being potentially limited in number of users, it is important to clarify the way that
Huawei measures the number of users in a cell/site:
Counter Counter
What is measured YL (both links) DL UL
family prefix Status
#users per cell L.Traffic.User Avg./Max. x x Active
DLData.Avg/ ULData.Avg/
#users with data in the buffers L.Traffic.User x Active
DLData.Max ULData.Max
#users transmitting/receiving
L.Traffic.User Data.Avg/Data.Max x x Active
data in a cell
# users in UL sync L.Traffic.User x x Ulsync.Avg Active
L.Traffic.Active To be
# active users User
Avg/Max DL.Avg/DL.Max UL.Avg/UL.Max
activated
#Carrier Aggregation users
PCell To be
having the local cell as primary L.Traffic.User x x
.DL.Avg/Max activated
serving cell
#Carrier Aggregation users
having the local cell as secondary SCell To be
L.Traffic.User x x
serving cell .DL.Avg/Max activated

General capacity/traffic cell counters

PRS LTE Radio I/F – cell traffic KPIs
Meas. Object KPI Index KPI name Formula Unit Aggregation SLAT
Avg DL User throughput L.Thrp.bits.DL / 10^6

R.C.T.L1
per cell /3600
[Mbps] T3W <12
Avg UL User throughput L.Thrp.bits.UL / 10^6
R.C.T.L2
per cell /3600
[Mbps] T3W <3
Cell
Avg. DL signalling L.Thrp.bits.DL.SRB /
R.C.T.L3 [Kbps] T3W tbc
throughput per cell 10^3 / 3600
Avg. DL signalling L.Thrp.bits.UL.SRB /

R.C.T.L4 [Kbps] T3W tbc
throughput per cell 10^3 / 3600
Upgrade path: R.C.T.L1or L2 violated 3/5 weeks  check the general Cell KPIs
(R.C.G.L1..7 defined further below) and if any of them are in violation then cell in
congestion  in such a case expand the number of carriers or add a sector or a site.
Meas
KPI Index KPI name Formula Unit Aggregation SLAT
Object
Average DL cell time L.Thrp.Time.Cell.DL/1000/3600/
R.C.U.L1 [%] T3W #N/A
utilisation (Avail_RBs*RB_Efficiency) (1)
Average UL cell time L.Thrp.Time.Cell.UL/1000/3600/
R.C.U.L2 [%] T3W #N/A
utilisation (Avail_RBs*RB_Efficiency) (1)
Cell
{L.Traffic.BCH.TB.bits} / 3600 /
R.C.T.L5 BCH Throughput [kbps] T3W #N/A
1000
PCH Throughput {L.Traffic.PCH.TB.bits} / 3600 /

R.C.T.L6 [kbps] T3W #N/A
[PCH_THROUGH] 1000
(1)  A quick procedure on how to obtain the “Avail_RBs” is explained in the
EE TEMPLATE FOOTER 11/09/2014
Cell PRB section below 34
LTE Radio I/F – cell traffic KPIs

PRS LTE Radio I/F – cell QCI traffic KPIs
Meas Aggreg
KPI Index KPI name Formula Unit SLAT
Obj ation
R.C.M.L1 QCI1 DL Mbps L.Thrp.bits.DL.PLMN.QCI.1(bit) / 10^6 / 3600 [Mbps] T3W tbc

Cell

Meas Aggreg
Obj ation
R.C.M.L11 QCI1 UL Mbps L.Thrp.bits.UL.PLMN.QCI.1(bit) / 10^6 / 3600 [Mbps] T3W tbc
Cell R.C.M.L16 QCI6 UL Mbps L.Thrp.bits.UL.PLMN.QCI.6(bit) / 10^6 / 3600 [Mbps] T3W tbc
Traffic volume of PDCP SDUs
R.C.T.L20 L.Thrp.bits.UE.UL.SmallPkt.PLMN / 10^6 / 3600 [Mbps] T3W tbc
scheduled for small packets
Traffic volume of PDCP SDUs
L.Thrp.Time.UE.UL.RmvSmallPkt.PLMN / 10^6 /
R.C.T.L21 scheduled for al traffic [Mbps] T3W tbc
3600
excluding small packets

LTE Radio I/F – cell QCI traffic KPIs

Meas Aggreg
Obj ation
Percentage of DL throughput Σ(L.Thrp.DL.BitRate.Samp.Index_k)/Σ(L.Thrp.DL.BitRa
R.C.K.D0 [%] T3W 5%
samples less than 2Mbps te.Samp.Index_j), k=0…2, j=0,9
Percentage of UL throughput Σ(L.Thrp.UL.BitRate.Samp.Index_k)/Σ(L.Thrp.UL.BitRa
R.C.K.U0 [%] T3W 5%
samples less than 1Mbps te.Samp.Index_j), k=0…2, j=0,9
Percentage of QCI1 throughput Σ(L.Thrp.DL.BitRate.QCI1.Samp.Index_k)/Σ(L.Thrp.DL.
R.C.K.1 [%] T3W 5%
samples less than 2Mbps BitRate.QCI1.Samp.Index_j), k=0…2, j=0,9

R.C.K.2 [%] T3W 5%

R.C.K.3 [%] T3W 5%
Cell R.C.K.4 [%] T3W 5%
R.C.K.5 [%] T3W 5%
R.C.K.6 [%] T3W 5%
Percentage of throughput QCI7 Σ(L.Thrp.DL.BitRate.QCI7.Samp.Index_k)/Σ(L.Thrp.DL.
R.C.K.7 [%] T3W 5%
R.C.K.8 [%] T3W 5%
R.C.K.9 [%] T3W 5%
Measure
ment KPI Index KPI name Formula Unit Aggregation SLAT
Object
Number of DRBs for services with L.Traffic.DRB.
R.C.T.L7 #DRBs T3W #N/A
the QCI of 1 in a cell QCI.1



Cell Number of DRBs for services with L.Traffic.DRB.


Meas KPI
KPI name Formula Unit Aggregation SLAT
Obj. Index
If(L.Traffic.DL.PktDelay.Num.QCI.1>10^3,
Avg. processing delay of DL PDCP SDUs
R.C.D.L1 L.Traffic.DL.PktDelay.Time.QCI.1/L.Traffi [ms] T3W 100
for DRB services with the QCI of 1 in a cell
c.DL.PktDelay.Num.QCI.1,0)
If(L.Traffic.DL.PktDelay.Num.QCI.2>10^3,L.Tr
R.C.D.L2 affic.DL.PktDelay.Time.QCI.2/L.Traffic.DL.Pkt [ms] T3W 100
for DRB services with the QCI of 2 in a cell Delay.Num.QCI.2,0)
Avg. processing delay of DL PDCP SDUs If(L.Traffic.DL.PktDelay.Num.QCI.4>10^3,L.Tr

R.C.D.L4 for DRB services with the QCI of 4 in a affic.DL.PktDelay.Time.QCI.4/L.Traffic.DL.Pkt [ms] T3W 100
cell Delay.Num.QCI.4,0)
Cell R.C.D.L5 affic.DL.PktDelay.Time.QCI.5/L.Traffic.DL.Pkt [ms] T3W 100



EE TEMPLATE FOOTER cell 11/09/2014 Delay.Num.QCI.9,0) 41
Meas KPI
KPI name Formula Unit Aggregation SLAT
Obj. Index
If(L.Traffic.DL.PktUuLoss.Tot.QCI.1>10^5,L.Traffic.UL.PktLo
R.C.P.1 DL.QCI1_Pkt_Loss [%] Weekly Max. 3%
ss.Loss.QCI.1/L.Traffic.DL.PktUuLoss.Tot.QCI.1,0)
Cell If(L.Traffic.DL.PktUuLoss.Tot.QCI.5>10^5,L.Traffic.UL.PktLo

LTE Radio I/F – cell traffic KPIs

PRS LTE Radio I/F – General cell congestion KPIs
Meas. Aggreg
Object
LBBP-c 240
Average in-sync
R.C.G.L1 L.Traffic.User. Ulsync.Avg #users T3W (1 LBBP-d1 240
Connected Users
LBBP-d2 240
LBBP-c 720
Average out-of-sync
R.C.G.L2 L.Traffic.User.Avg - L.Traffic.User. Ulsync.Avg #users T3W (1 LBBP-d1 1440
Connected Users
LBBP-d2 1440
E-RAB resource L.E-RAB.FailEst.NoRadioRes/L.E-RAB.AttEst x Wk

R.C.G.L3 [%] 0.2%
congestion rate 100% max
Radio
cell E-RAB resource
L.E-RAB.FailModify.NoRadioRes/L.E-RAB.AttEst x Wk.
R.C.G.L4 modification congestion [%] 0.2%
100% max
rate
E-RRC resource L.RRC.SetupFail.ResFail/L.RRC.ConnReq.Att x Wk.

R.C.G.L5 [%] 0.2%
congestion rate 100% Max
RRC failure rate due to

R.C.G.L6 L.RRC.SetupFail.ResFail [%] T3W 1%
no resources available
Number of UE context
If(L.UECNTX.AttEst>100,L.UECNTX.FailEst. Wk
R.C.G.L7 failure rate due to Radio [%] 1%
Congestion NoRadioRes/L.UECNTX.AttEst,0) Max
44
Upgrade path:
R.C.G.L1,2 violated 3/5 weeks  follow the Huawei recommended LBBP expansion procedure
mentioned earlier in this chapter, in section “b. Radio Site aspects”; (1)  T3W = Top 3 Weekly value,
R.C.G.L3-7 violated 3/5 weeks  Correlate with other violated Cell defined in the Introduction
KPIs (listed below) and optimise as per each specific table at the beginning of this
11/09/2014
EE TEMPLATE FOOTER
violated KPI’s “expansion steps” chapter
PRS LTE Radio I/F – cell traffic KPIs
Measur
ement KPI Index KPI name Formula Unit Aggregation SLAT
Object
If{(L.Traffic.DL.SCH.QPSK.TB+L.Traffic.DL.SCH.QPS
DL QPSK TB(1) K.TB.Retrans)>10^5,L.Traffic.DL.SCH.QPSK.TB.Retr
R.C.Q.L1 ans/(L.Traffic.DL.SCH.QPSK.TB+L.Traffic.DL.SCH.Q
[%] T3W 10%
retransmission rate
PSK.TB.Retrans),0}
If{(L.Traffic.DL.SCH.16QAM.TB+L.Traffic.DL.S
DL 16QAM TB(1) CH.16QAM.TB.Retrans)>10^3,L.Traffic.DL.SCH
R.C.Q.L2 [%] T3W 30%
retransmission rate .16QAM.TB.Retrans/(L.Traffic.DL.SCH.16QAM.
TB+L.Traffic.DL.SCH.16QAM.TB.Retrans),0}
If{(L.Traffic.DL.SCH.64QAM.TB+L.Traffic.DL.S
DL 64QAM TB(1) CH.64QAM.TB.Retrans)>10^2,L.Traffic.DL.SC
Cell R.C.Q.L3 [%] T3W 80%
retransmission rate H.64QAM.TB.Retrans/(L.Traffic.DL.SCH.64QA
M.TB+L.Traffic.DL.SCH.64QAM.TB.Retrans),0}
If{(L.Traffic.UL.SCH.QPSK.TB+L.Traffic.UL.SCH.QPS
UL QPSK TB(1) K.TB.Retrans)>10^5,L.Traffic.UL.SCH.QPSK.TB.Retr
R.C.Q.L4 ans/(L.Traffic.UL.SCH.QPSK.TB+L.Traffic.UL.SCH.Q
[%] T3W 10%
retransmission rate
PSK.TB.Retrans),0}
If{(L.Traffic.UL.SCH.16QAM.TB+L.Traffic.UL.S
UL 16QAM TB(1) CH.16QAM.TB.Retrans)>10^3,L.Traffic.UL.SCH
R.C.Q.L5 [%] T3W 30%
retransmission rate .16QAM.TB.Retrans/(L.Traffic.UL.SCH.16QAM.
TB+L.Traffic.UL.SCH.16QAM.TB.Retrans),0}
(1) TB= MAC Transmission Block

c1. PRACH

LTE Radio I/F – 1. PRACH (UL) - general introduction
• Resource that gets congested: A cell can use a maximum of 64 random access preambles, with indexes of 0 to 63;
• RACH preambles are divided into:
• random preambles (contention random access): contention-based RACH  for initial access or UL
synchronization; there are two classes: A and B; the UE chooses group B if:
• Group B exists;
• If size(Msg3)>MessageSizeGroupA param;
• Pathloss(UE) ≤ Threshold
• dedicated preambles (non-contention random access): non-contention RACH  for Hos . If dedicated
RACHs are insufficient, the eNodeB has to instruct the UE to initiate contention-based random access
• RACH resources are configured semi-statically  if the “RACH optimization feature” is NOT active, the Operator
must dimension/define the resources;
• RACH resources need to be adjusted in scenarios such as:

• #HO attempts increases  dedicated preambles may be insufficient and UEs need to initiate contention-based random
access, which increases handover delays.
• # random access attempts increases, random preambles or
PRACHs may be insufficient and preamble collisions increase
on PRACHs, which increases access delays.
0
The default value of sizeOfRA-PreamblesGroupA is 28
The default value ofnumberOfRA-Preambles is 52
• RACH is used for:
27 Contention for
• UL synchronization for new UL data;
• UL synchronization for new DL data (contention-free); access preambles
• HO (contention-free);
51
• TA update in RRC_Connected;
• Transition from RRC_IDLE to RRC_CONNECTED
• Recovering from link failure 63 Non-contention (HO)
11/09/2014
• An UL synchronized UE sends a SR message (Scheduling Request) 47
LTE Radio I/F – 1. PRACH (UL) – capacity SLAs
• PRACH slots in LTE follow a slotted Aloha protocol;
• There are 64 PRACH sequences per cell (as per previous slide), of which 12 can be used for HOs and
UL re-synchronization due to incoming DL data;
• If the 12 HO PRACH sequences are being exhausted, the system will allocate contention-based
PRACH sequences, leading to possible HO success rate degradation  hence the need to define an
upper limit for the dedicated #PRACH attempts per hour and also for the total PRACH attempts per
hour as well;
• Slotted Aloha blocking probability – theoretical value follows the formula:
𝑃𝑏𝑙𝑜𝑐𝑘 = 1 - 𝑒 −𝐺 [%] where G = Offered traffic (Erlangs or Packets/”packet time”)
• In this calculation G is the number of PRACH attempts per 5 MHz and 10ms 
𝐺
(− )
𝑃𝑏𝑙𝑜𝑐𝑘 𝐿𝑇𝐸 𝑃𝑅𝐴𝐶𝐻 = 1 -𝑒 #𝑃𝑅𝐴𝐶𝐻_𝑆𝑒𝑞
[%](1)
• For a Pblock_LTE_PRACH_MAX (for example 1%)  GMAX = - #PRACH_Seq X (1 – Pblock_LTE_PRACH_MAX )
• In the table below we present the equivalent SLA for the #PRACH_atts per hour, based on 0.1%
PRACH total blocking probability and also the probability that the HOs and DL Data re-
synchronizations are being served by contention-based PRACH of less than 0.1%
(1) = According to the PRACH

dimensioning presented in Ref.
[1], par. 17.2.5 PRACH Resource
Configurations
LTE Radio I/F – 1. PRACH (UL) - Huawei features and parameters
• In the Huawei domain, the eNB is capable of a feature called “automatic
RACH optimisation”, which has two parts:
• RACH Resource adjustment:
• As defined in 3GPP specifications, RACH resources for a cell consist of preambles and PRACH
resources;
• Preambles are further classified as dedicated preambles and random access preambles;
• In Huawei RACH Resource Adjustment there are two RACH optimization mechanisms:
• preamble groups  the eNB adjusts the #preambles in each group;
• PRACH configurations  the eNB adjusts the #PRACHs allocated each second
• Adaptive backoff:
• Adaptive backoff enables the eNodeB to specify the backoff period for UEs based on the
contention-based random access load;
• In EE’s network, currently RACH Algorithm Switch has a value of 11111, meaning:
• RachAdjSwitch = 1  RACH Resource Adjustment active (see above);
• UnsyncRaSwitch = 1  if 1, UEs will use the non-contention RACH during UL synchro;
• BackoffSwitch = 1  backoff control enabled;
• HoRaSwitch = 1  if 1, UEs will use the non-contention RACH during HOs;
• MaskIdxSwitch = 1  if 1, allow the reuse of dedicated preambles, based on MaskIndex
parameter; if 0, only 1 UE per dedicated pre-amble.

LTE Radio I/F – 1. PRACH (UL) – Huawei Counters

PRS
Meas.
LTE Radio I/F – 1. PRACH - congestion KPIs Aggr
Object eg.
PRACH Random pool preamble

R.C.PR.L1 usage
(L.RA.GrpA.Att + L.RA.GrpB.Att)/3600/100 x 100% [%] T3W 75%
PRACH Dedicated pool preamble

Cell R.C.PR.L2 usage
L.RA.Dedicate.Att/3600/100 x 100% [%] T3W 75%
(L.RA.GrpA.Att + L.RA.GrpB.Att + L.RA.Dedicate.Att) /

R.C.PR.L3 Total PRACH load [%] T3W 80%
90000 (1)
R.C.PR.L4 Non-contention PRACH load (L.RA.Dedicate.Att) / 18000 (1) [%] T3W 80%
Upgrade option: Because all the RACH adaptive feature options are currently activated, the only option for
upgrade in this case is traffic rebalancing, followed by a carrier upgrade or a sector split.
11/09/2014
(1)  please check the justification for the 90000 and 18000 term in the PRACH capacity calculation in the previous slide)
c2. PRB resources

PRS LTE Radio I/F –2. PRBs (DL/UL) - congestion KPIs
Meas. Aggreg
Object
R.C.R.L1 DL Available PRBs L.ChMeas.PRB.DL.Avail #N/A Max #N/A

R.C.R.L2 UL Available PRBs L.ChMeas.PRB.UL.Avail #N/A Max #N/A
Radio R.C.R.L3 DL 20MHz PRB Usage 100 x (L.ChMeas.PRB.DL.Used.Avg / 100) [%] T3W 80%
cell
R.C.R.L4 UL 20MHz PRB Usage 100 x (L.ChMeas.PRB.UL.Used.Avg / 100) [%] T3W 80%
DL user-perceived
R.C.R.L5 L.Thrp.bits.DL/L.Thrp.Time.DL/1000 [Mbps] T3W 2
throughput
Upgrade path:
If R.C.R.L1 violated 3/5 weeks and T3W “DL user-perceived throughput” <2Mbps 3/5
weeks  add Radio carrier, then if this is not possible, then split the sector;
How to obtain the Available PRBs until the corresponding counters will be activated: use the CPM tool
https://cpm.orange.co.uk/download/LTE/4GCPMDat.ZIP (updated daily) ==> in the table H_CELL in
11/09/2014 each of the databases in this zip, the two columns to extract are CELLNAME and DLBANDWIDTH, 53
EE TEMPLATE FOOTER which have to be correlated with the “DLBANDIWTDH” table to get the value in MHz.
c3. SRS resources

LTE Radio I/F – 3. SRS (UL) resources - general introduction
• Sounding Reference Signal (SRS) is a resource allocated by the eNB, and is used to provide information on uplink
channel quality on a wider bwidth than the PUSCH transmission or when terminal has no transmissions on PUSCH;
• SRS is used for UL channel estimation, for closed loop precoding, for UL scheduling, link adaptation, UL power
control, antenna selection and maintenance of synchronization and the DL transmission.
• The SRS requires that each tx. antenna transmits SRSs with
a period shorter than the channel coherence time and at a
bandwidth significantly larger than the channel coherence bandwidth
• UE uses every second sub-carrier for transmitting reference
signal;
• There are several ways to multiplex SRS users in UL:
A. In time in different subframes;
B. In frequency  the PRB possible sizes (in PRBs) follow
a tree structure (see bottom line figure); the frequency
hopping pattern follows also the tree structure;
C. Cyclic shifts (ZC sequences)  max. 8 cyclic shifts; the
sequence group configured for PUCCH is used also
for SRS. • SRS transmission can also hop in frequency; frequency
D. Transmission comb  2 combs available (even or odd hopping can also be limited to a certain portion of the
subcarriers); system bandwidth, which is beneficial for inter-cell
• SRS sequence lengths are multiples of 24, or, interference coordination;
correspondingly, SRS bandwidths are multiples of 4 RBs;
• In addition to the terminal-specific SRS configuration, the
• An SRS transmission can be a single transmission or
cell-specific SRS configuration defines the sub-frames,
transmissions can be periodic with the period ranging from 2
which can contain SRS transmissions as well as the set
ms to 320 ms;
of SRS bandwidths available in the cell
• Up to four different SRS bandwidth options available,
depending on the system bandwidth and cell configuration; FREQUENCY
TIME
LTE Radio I/F – 3. SRS (UL) resources –multiplexing aspects
FREQUENCY
• Periodic and aperiodic SRS transmission
share a common set of cell-
specific SRS resources (subframe
configuration period, subframe offset, and
maximum SRS bandwidth);
• Different sets of UE-specific sounding

parameters are independently allocated for
periodic and aperiodic SRS transmission
including transmission bandwidth,
periodicity, frequency comb pattern, and TIME
cyclic shift;
Fig. 1 Typical example of SRS multiplexing
• The aperiodic SRS-related signalling bits are
piggybacked in either a downlink
assignment (scheduling downlink data
transmission) or an uplink grant (scheduling
uplink data transmission), in order to reduce
signalling overhead
• If SRS resources are insufficient, the

eNodeB cannot obtain the UE measurement
information, leading to a waste of air
interface resources Fig. 2 Another example of SRS multiplexing
• Huawei recommends, if the value of SoundingRS algorithm switch is SrsSubframeRecfSwitch:On,
that: SRS resource usage = L.Traffic.User.Ulsync.Avg/(Maximum #UEs supported by SRS resources) x 100% ≤ 60%
11/09/2014 56
LST SRSADAPTIVECFG LST CELLALGOSWITCH
LTE Radio I/F – 3. SRS (UL) - Huawei features and parameters
• An SRS subframe configuration index
defines a subframe period and a
subframe offset;
• If SRSs are transmitted every T ms, T
is the SRS subframe period;
• The SrsSubframeCfg parameter
specifies the index for a cell -specific
SRS subframe configuration;
• The eNodeB can dynamically adjust
SRS subframes based on cell loads;
• If the SrsAlgoSwitch parameter is set
to SrsSubframeRecfSwitch -1, the
eNodeB adaptively adjusts SRS
subframe configuration based on cell
loads.

PRS
LTE Radio I/F – 3. SRS resources - congestion KPIs
Meas. Aggr
Object eg.
R.C.S.L1 SRS resource Usage 100 x L.Traffic.User.Ulsync.Avg/400 [%] T3W 60%

Cell
RRC failure rate due to no SRS L.RRC.SetupFail.ResFail.SRS/L.RRC.ConnReq.At
R.C.S.L2 resources
[%] T3W 1%
t, where L.RRC.ConnReq.Att>50
Upgrade path: If the SRS resource usage exceeds 75%, check the Huawei SRS parameters, as per the previous slide.
Increase the SRS resource configuration until it reaches the maximum SRS subframe configuration. Alternatively, it
is recommended that SRS resource adaptation be enabled.
If the SRS resource adaptation already enabled, then try to rebalance the traffic between sectors, if this is difficult
to achieve, the next step is a carrier addition; if this is not possible due to the maximum number of carriers is already
11/09/2014
achieved, the next steps are sector split followed by new site.
c4. PUCCH resources

LTE Radio I/F – 4. PUCCH (UL) resources – general aspects
• PUCCH carries:
• SRI  Scheduling Resource Indicator SRI (the UE will send a SRI when has data to tx. in UL); SRIs and ACKs
can share the same RBs; in Huawei there is an adaptive SRI allocation algorithm (bottom center figure);
• CQI/PMI/RI  Feedback of DL channel quality indication (CQI, Precoding Matrix Ind. and Rank Ind.);
• ACK  ACK to DL HARQ (indicating a PDSCH decoding success) - A UE uses an ACK to indicate PDSCH
decoding success. If the UE successfully receives data from the eNodeB, it sends an ACK to the eNodeB.;
• PUCCHs occupy resource blocks (RBs) at two ends of an uplink cell bandwidth and use frequency hopping to
achieve frequency diversity gain (right figure);
time
1. SRI:
freq
• If SRI resources are insufficient or limited, the eNodeB selects a longer

SRI period (e.g. 20 ms) for UEs, which results in larger uplink scheduling
delays for these UEs;
• SRIs and ACKs can share the same RBs;
• SRIs occupy PUCCH resources, and their occupancy is represented by SRI loads

LTE Radio I/F – 4. PUCCH (UL) resources – general aspects (cont.)
2. CQI/PMI/RI:
• A UE reports a CQI to the eNodeB periodically, or in event-triggered mode. Event-triggered reporting takes
precedence over periodic reporting if both are configured;
• CQIs reported in event-triggered mode are sent to the eNodeB over a PUSCH;
• If CQI resources are insufficient or limited, the eNodeB selects a CQI period of 40 ms for UEs, which results in
less accurate CQIs and lower spectrum efficiency. In addition, these UEs cannot be chosen to perform
frequency selective scheduling.
• PMI  To avoid interference between spatial channels, codewords are mapped onto different layers, and
layered data is multiplied by a precoding matrix and mapped onto different antenna ports for transmission.
• RI indicates the rank of a spatial channel matrix
• The eNodeB dynamically adjusts PUCCH resources based

on PUCCH resource load. By reducing the overhead, more
resources can be used for PUSCH transmissions. By
increasing the PUCCH resources, more UEs can be
admitted;

LTE Radio I/F – 4. PUCCH (UL) resources – Huawei features and parameters
• Two features that impact PUCCH allocation:
1. Dynamic PUCCH resource adjustment: the eNodeB dynamically adjusts PUCCH resources
based on PUCCH resource load
2. SRI adaptation: The eNodeB by default adaptively adjusts an
SRI period based on SRI loads;.
• If the GLOBALPROCSWITCH.SriAdaptiveSwitchparameter is set to ON(On), the eNodeB adaptively

adjusts the SRI period for a newly admitted UE based on SRI loads.
• SRI loads are classified into low, medium, and high;

• A higher SRI load results in a larger SRI period set for a newly
admitted UE;
• For medium SRI loads, set an SRI period for an admitted UE
based on the QCI of a service;
• For low SRI loads, set SRI periods for all admitted UEs to 5 ms;
• The CELLPUCCHALGO.SriLowLoadThdparameter (10 in EE)

specifies the threshold for the low load state of SRI
resources. The value range is 0 to 50, indicating the
number of UEs that can be admitted under the low load
state. For the other two stages (medium and high)
Huawei has two specific tables for the SRI period [ms];
LTE Radio I/F – 4. PUCCH (UL) resources – Huawei counters

PRS LTE Radio I/F – 4. PUCCH Traffic/congestion KPIs
Meas. Aggr
Object eg.
RRC failure rate due to no PUCCH L.RRC.SetupFail.ResFail.PUCCH/L.RRC.ConnReq.At

R.C.PU.L1 [%] T3W 1%
resources t, where L.RRC.ConnReq.Att>50
Cell
R.C.PU.L2 PUCCH resource usage 100 X L.Traffic.User.Ulsync.Avg/400 [%] T3W 75%
R.C.PU.L3 PUCCH Avg PRBs L.ChMeas.PRB.PUCCH.Avg #PRBs T3W #N/A
If the PUCCH resource usage exceeds 75%, increase PUCCH resource configuration if the dynamic
PUCCH algorithm not active; do this until the maximum number of CQI RBs and SRI resources are
being used. Alternatively, it is recommended that PUCCH resource adjustment be enabled. In this
case, if the PUCCH resource usage is still high, try to implement a traffic rebalancing between the
sectors, followed by a carrier addition/sector split and ultimately a new site.
11/09/2014
c5. PDCCH resources

LTE Radio I/F – 5. PDCCH resources – general aspects
• Downlink control channels are carried over the Physical Downlink Control Channel
(PDCCH) and they contain control information from the MAC layer, including
downlink control information (DCI), Control Format Indicator (CFI), and H-ARQ Indicator
(HI).
• The Physical Downlink Control Channel (PDCCH) carries the DCI info:
1. DL grant: PDSCH resource ind., MCS, HARQ, PUCCH pwr ctrol.
2. UL grant: PUSCH resource ind., MCS, HARQ, PUSCH pwr ctrol.
3. Power control  a supplemental group of PUSCH pwr ctrol.
• Can be power-controlled;
• Within the 1 ms sub-frame, only the first 0.5 ms slot contains PDCCH;
• A control channel element (CCE) that consists of nine resource element groups is the
minimum unit of transmission for PDCCH;
PDCCH Aggregation levels:
• Downlink data transmission can be scheduled
dynamically every sub-frame or semi-
1,2,4 or 8 CCEs = 1 PDCHs 1 CCE = 9 X REGs
1 RE
persistently until the resource is released;.
• For dynamically scheduled data transmission,
the scheduling assignment is given via
downlink control information
100 RBs
(DCI) format1/1A/1B/2/2A/2B
PCFICH + PHICH + PDCCH
In slots 0…2
Uplink CCEs transmit uplink 1 ms
scheduling grants, and downlink CCEs
transmit downlink scheduling grants.
EE TEMPLATE FOOTER 11/09/2014 1 REG group = 4 X REs 66
LTE Radio I/F – 5. PDCCH resources – Huawei features and parameters
The two Huawei PDCCH rules:
• PDDCH Symbol utilisation: If the values of the first two counters L.ChMeas.PDCCH.SymNum.3 and 4
are greater than the values of L.ChMeas.PDCCH.SymNum.1 and 2, the PDCCH symbol usage is high
• DL CCE Utilisation: If the value of L.ChMeas.CCE.DLUsed is greater than the value of

L.ChMeas.CCE.ULUsed, the PDCCH downlink CCE usage is high.
Huawei PDCCH related algorithms:
1. PDCCH symbol adaptation (enabled by default)  adjust the

number of symbols occupied by a PDCCH based on the number
of required CCEs;
2. PDDCH aggregation level adaptation  select an appropriate

PDCCH aggregation level based on the channel quality CQI;
3. Closed-loop adjustment of PDCCH aggregation level 

dynamically adjusts the PDCCH aggregation level based on the
PDCCH BLER, in addition to the adjusted CQI;
4. CCE Ratio adaptation: the CCE ratio is the ratio of the number
of uplink CCEs to the number of downlink CCEs  the eNodeB
dynamically adjusts the CCE ratio based on the usage of uplink
and downlink CCEs.

LTE Radio I/F – 5. PDCCH resources – Huawei counters
Needs activation

PRS LTE Radio I/F – cell Traffic/congestion KPIs
Meas. Aggr
Object eg.
If(L.ChMeas.CCE.ULUsed>10^6,
R.C.C.L1 DL CCE resource usage L.ChMeas.CCE.DLUsed/L.ChMeas.CCE.ULUsed,0)
[%] T3W 90%
Cell
(L.ChMeas.PDCCH.SymNum.3+L.ChMeas.PDCCH.Sym
R.C.C.L2 PDCCH Symbol utilisation Num.4)/(L.ChMeas.PDCCH.SymNum.1+L.ChMeas.PDC [%] T3W 90%
CH.SymNum.2)
Upgrade path: If the “DL CCE resource usage” or “PDCCH Symbol utilisation” T3W values exceed 90%, check the
Huawei PDCCH parameters, as per the previous slide.
If any of the adaptation features is not active, activate it, then try to rebalance the traffic between neighbour
sectors, if this is difficult to achieve, the next step is a carrier addition; if this is not possible due to the maximum
11/09/2014
number of carriers already achieved, the next steps are sector split followed by new site.
c6. Paging resources

LTE Radio I/F – 6. Paging resources – general aspects
• The eNodeB and BTS3202E or BTS3203E LTE can process a maximum of 750 and 500
paging messages per second, respectively; this limit for a BTS3900 is (tbc???)
• If the number of paging messages exceeds that capacity, paging messages sent from the eNodeB
to UEs may be discarded;
• The eNBs do not use the Paging procedure to page a UE in RRC_CONNECTED;
• UEs in RRC_CONNECTED as well as UEs in RRC_IDLE are required to monitor Paging messages, in
order to be aware of changes in system information;
• Additionally, the Paging procedure can be used to indicate warning notifications broadcast on
the BCCH;

LTE Radio I/F – 6. Paging resources – Huawei counters

PRS LTE Radio I/F – cell traffic/congestion KPIs
Measure Needs activation

ment KPI Index KPI name Formula Unit Aggregation SLAT
Object
100 X {[L.Paging.S1.Rx /
R.C.A.L1 S1 Paging Utilisation
3600] / 750}
[%] T3W 60%
#Discarded Paging msg due L.Paging.Dis.PchCo

Cell R.C.A.L2
to congestion ng
#msg Wk Max >0
100 X {[L.Paging.UU.Att /
R.C.A.L3 Uu Paging Utilisation
3600] / 750}
[%] T3W 60%
Upgrade path: If, for more than 30% cells in a TA, (either of the two R.C.A.L1 or R.C.A.L3 are violated 3/5 weeks) and
R.C.A.L2 also violated 3/5 weeks, then re-dimension the Tracking Area. Otherwise, if less than 30% cells in violation
then reconfigure the Paging parameters to reduce the Paging Traffic (measured through the R.C.T.L6 KPI defined in
the Cell/General KPIs section.
c7. Flow control

Meas.
Obj.
#Discarded Uu paging
R.C.F.H1
messages due to Flow ctrol
L.Paging.Dis.Num #msg Wk Max >0
Number of times uplink

R.C.F.H2 L.LC.ULCong.Num.Cell [%] Wk Max >0
congestion occurs in a cell(1)
Number of times downlink

R.C.F.H3 L.LC.DLCong.Num.Cell [%] Wk Max >0
congestion occurs in a cell(1)
If(L.RRC.ConnReq.Att>100,L.RRC.Co
RRC failure rate due to MME
Cell R.C.F.H4 flow control
nnReq.Msg.disc.FlowCtrl/L.RRC.Con [%] T3W 0.5%
nReq.Att,0)
If(L.RRC.ConnReq.Att>100,L.RRC.Co
RRC discard failure rate due to
R.C.F.H5 MME flow control
nnReq.Msg.disc.FlowCtrl/L.RRC.Con [%] T3W 0.5%
nReq.Att,0)
Discards of HO Req due to MME

R.C.F.H6 flow control
L.HHO.PrepAttIn.disc.FlowCtrl # Wk. Max >0
#HO preparation failures due to

R.C.F.H7 MME flow control
L.HHO.Prep.FailIn.FlowCtrl # Wk. Max >0
Upgrade path: If, for more than 30% cells in a TA, (either of the R.C.F.H1 to R.C.F.H7 are being violated 3/5 weeks),
then the system is in the so-called “flow controlled” region, trying to reduce the load in different domains by
artificially reducing the traffic flows. In such a case a more thorough analyses is needed, as this could be a major sign
of stress for an entire area of a network.
(1) This KPI only available when the “Congestion control” feature activated in the cell
Meas.
Obj.
R.C.FD.H1 Discarded QCI1 SDUs due to
L.PDCP.Tx.Disc.Trf.SDU.QCI.1 #SDUs T3W 10^4
user-plane Flow ctrol
Cell L.PDCP.Tx.Disc.Trf.SDU.QCI.5 #SDUs T3W 10^4

4. LTE Transport capacity

4a. LTE Transport - Backhaul capacity

LTE Transport capacity – Backhaul capacity
• To be as clear as possible, the backhaul capacity has been split
in two (see figure below):
• eNB side
• Network side
Web
SGW
eNB
CPCN/
GRN/
core PE MSN PE
MPLS Switch sites MSN
MME network network
IIa. Connectivity
domain IIb. Backhaul domain
eNB
PKI
II. LTE Transport I. Radio
DHCP I/F
eNB domain
Backhaul
III. Infrastructure side
network
domain side
4a1. LTE Transport - eNB side

a. eNB transport

LTE Transport - eNB side – eNB Transmission modules
• Two type of eNB transport modules: LMPT and UMPT
LMPT UMPT
• LMPT card supports the following features • UMPTa6 card supports the following features:
• 2x Electrical 2x Optical Ethernet ports • 1x Electrical & 1x Optical ports for
for transmission (all FE/GE). transmission (all FE/GE).
• LMPT card allow access to Local • 4 x E1 ports
management terminal via ETH port. • Inbuilt GPS receiver, a GPS antenna is
• Key capacity limits of 1xLMPT are following required to use this function
• 1800 RRC In-sync RRC connected • LMPT card allow access to Local
users management terminal via adaptor connected
• 3600 RRC Out-of-sync RRC to USB port.
connected users
• DL+UL throughput 750Mbps with • Key capacity limits of 1xUMPT are
IPSec enabled • 3600 RRC In-sync RRC connected users
• 7200 RRC Out-of-sync RRC connected users
• DL+UL Throughput 1.5Gbps with IPSec enabled
LTE - Backhaul eNB side –– QoS aspects
Upstream direction, the Huawei eNodeB has three QoS features:
• Line rate policer  can be applied to the backhaul interface;
• TRG shaper  shaping of all user plane traffic on IP path towards S-GW;
• Traffic Control Switch (TCSW) allows the eNodeB to signal congestion back to PDCP
thus providing smoother (less spikey) TCP/UDP flows when multiple user streams exist.
• The CBS, CIR, PIR, and PBS settings are of no interest and configured such as to satisfy the
configuration checking process on the eNB.
• Shaping action is set to single rate rather than dual rate in the TRG parameters.
• If multiple EE eNBs are provisioned at a non dependency site, the QoS variables need to be
adapted so that the aggregate sum of all the deployed eNBs does not exceed the provisioned
backhaul bandwidth for LTE. As an example; if two eNBs are being deployed on a BT MEAS 300
non split site then the line rate and TRG TXBW should be configured to half the 90Mbps on each
eNB. In addition the TRG committed and burst parameters should be adjusted accordingly (
TXCBS = 45 002 Kbps, TXCIR = 45 001 Kbps, TXPIR = 45 002 Kbps, TXPBS = 45 002 Kbps)

Name Type of Traffic PHB Deci
mal
Value
LTE - Backhaul eNB side –– QoS aspects QCI 1
QCI 2
Conversational Voice
Conversational Video (Live

EF
AF41
46
34
Streaming)
QCI 3 Real-time gaming AF41 34
QCI 4 Non-conversational video AF21 18

(buffered streaming)
QCI 5 IMS signaling CS5 40
QCI 6 Video (buffered streaming), AF21 18
TCP based services
QCI 7 Voice, video (live AF11 10
streaming), interactive
gaming
QCI 8 Premium bearer for video, AF11 10
TCP based services for
premium subs
QCI 9 Default bearer for video, BE 0
TCP based-services for non-
premium subs.
LTE LTE Control Plane CS5 40
CP
LTE LTE Management Plane CS2 16
MP

LTE Transport - eNB side – protocol stack and tx. modules
The eNB side is treated in this document as having 4 layers(1)
Layer 4
Layer 3
Layer 2
Layer 1
(1) For brevity, it is considered that GTP-U is a transport

layer, however this is actually an Application Layer
(2) A GTP-U (GPRS Tunnelling Protocol - User plane) data path is a unidirectional connection between two GTP-U
entities which are identified by their IP destination address
LTE Transport eNB side(1) Layer 4 – GTP-U tunnels and SCTP links
• A “GTP-U tunnel” is a directional connection between two IP addresses;
• An “SCTP link” is a connection between two SCTP endpoints;
• An “S1 transport bearer” is a pair of two opposite directed bi-directional GTP-U tunnels between eNB and S-GW. SGW1
• An “X2 transport bearer” consists of only one unidirectional GTP-U tunnel
• An “SCTP endpoint” is a pair of (IP address, SCTP port); At any given time each node can set up exactly one “SCTP association”
• A pair of destination and source IP addresses is called an “SCTP transport path”
Unidirectional Destination IP Address SGW1
GTP-U tunnel over X2
eNB B X2 transport bearer
SGW1
S1 transport bearer
SCTP link Pair of GTP-U tunnels
over X2
over S1 transport bearer
MME1
SCTP link over S1 MME1

eNB A
SCTP endpoint
MME1
Source IP Address SCTP endpoint

LTE Transport eNB side(1) Layer 4 – congestion points
GTP-U data paths are established on the S1 interface between S-GW and eNB, and on SGW1
the X2 interface between two adjacent eNBs
IP_Path1
SGW2
eNodeB 2 Physical IP Addr
IP_Path4 IP_Path2
UP IP addr
CP IP addr 3. IP PATH
MH IP addr IP-Sec
X2 UP
SGW3
O&M IP addr X2 CP S1 UP IP_Path3
SCTP_Assoc 4
Sec-GW
2. GTP-U
tunnel Physical IP Addr MME1
1. S1 link UP IP addr IPSec Tunnel
4. X2 link CP IP addr SCTP_Assoc 1
IP-Sec
MH IP addr
O&M IP addr SCTP_Assoc 2
MME2
eNodeB 1
SCTP_Assoc 3
MME3
LTE transport domain – Backhaul eNB side S1 CP
(1) This document does not include the MGW and SGW side
EE TEMPLATE FOOTER 11/09/2014 of the LTE Transport 87
LTE Transport eNB side Layer 4 – explanations
• The diagram presented in the previous slide is a simplified diagram of the LTE transport system,
meant to explain mainly the Capacity aspects linked to the LTE BAU Capacity process;
• Each site (eNodeB) is linked to the Sec-GW by an IP-Sec tunnel;
• Each eNodeB has a physical IP address;
• Internally, on the Transport side, each eNodeB has two parts: User Plane (UP) and Control Plane
(CP)
• In addition to the physical IP address, the eNodeB has an additional set of an “end2end IP logical”
addresses: a UP IP address, a CP address, an O&M address and a Multi-homing CP IP address (the
last two are not presented in the above diagram for simplicity);
• From the UP IP logical address, there are several IP Paths “connections” (similar to user-plane
pseudo-wires) which connect the eNodeB to the three SGWs and the adjacent eNodeBs (linked
through X2 UP);
• From the CP IP logical address, there are several SCTP links (similar to signaling pseudo-wires)
which connect the eNodeB signaling IP address to the MMEs and the adjacent eNodeBs (linked
through S1 UP);
• The following elements are being reported in the PRS:
S1 links, GTP-U tunnels, IPPATH UDP links, X2 links, SCTP associations, ETH ports;
• Each of these elements are potential bottlenecks on the eNodeB “Transport”
PRS LTE Transport eNB side Layer 4 – capacity KPIs
Meas. KPI Aggrega
KPI name Formula Unit SLAT
Object Index tion
If({L.Sig.S1.SendSetup.Att}>0,100 X Count of
T.B.N.L1 S1 Setup Failure Rate (1-{L.Sig.S1.SendSetup.Succ } / [%] (samples >5
Cell {L.Sig.S1.SendSetup.Att}),0) >80%)
T.B.N.L2 S1 Setup Attempts {L.Sig.S1.SendSetup.Att} #atts. Total wk >0
E-RAB setup failure rate due to faults

T.B.S.L1 L.E-RAB.FailEst.TNL #setups Total wk >0
at the transport network layer
T.B.S.L2 GTP-U Rx Throughput(1) {Gtpu.RxBytes} x 8 /3600 [Mbps] T3W #N/A
T.B.S.L3 GTP-U Tx Throughput(1) {Gtpu.TxBytes} x 8 /3600 / 10^6 [Mbps] T3W #N/A
If(VS.Gtpu.RxBytes>10^6,VS.Gtpu.R
T.B.S.L4 GTP-U Receive Drop Rate [%] Wk. Max 0.1%
Site xDropBytes/VS.Gtpu.RxBytes,0)
If(VS.Gtpu.RxPkts>10^5,VS.Gtpu.Tx Weekly >150% or

T.B.S.L5 GTP-U UL/DL Packet Ratio [%]
Pkts / VS.Gtpu.RxPkts,0) Average <50%
T.B.S.L6 X2 Local Link Setup Attempts Sum(L.Sig.X2.SendSetup.Att #atts Total wk >0
If(L.Sig.X2.SendSetup.Att>0,100 X (1- Count of

T.B.S.L7 X2 Local Link Setup Failure Rate {L.Sig.X2.SendSetup.Succ} / [%] (samples 5
{L.Sig.X2.SendSetup.Att}),0) >80%)
T.B.S.L8 eNB X2 Utilisation #X2-links / 63 [%] #N/A 80%
• If T.B.N.1..2, T.B.S.6..7  under normal circumstances they are 0, if >0 then check the Transport section of that site, as follows:
Check the T.B.N layers 2 and 3 KPIs (defined further below) to identify the transport segment under pressure - if layer 2 counters violated
the the segment between the eNB and the MSN PE is congested; if layer 3 KPIs OK but the layer 3 KPIs are in violation then the problem resides in one of
the routers across the path (CX600, AR or Sec_GW)  in such a case check the relevant KPIs (defined further below in the corresponding sections) and
identify the router under congestion.
• If T.B.S.4 in violation  the eNB cannot serve the traffic as fast as the S-GW is sending the data  check the Radio I/F congestion;
• If T.B.S.5 in violation  there is a traffic imbalance DL vs. UL at the transport layer  check the Layer 3 Transport KPIs (defined further below) for a
routing problem. (1) Technically, the GTP-U layer belongs to the Application Layer
LTE Transport eNB side – transport capacity KPIs reflected in cell

LTE Transport eNB side – transport capacity KPIs reflected in cell

PRS LTE Transport eNB side Layer 4 – capacity KPIs (cont.)
Meas. KPI Aggreg

KPI name Formula Unit SLAT
Object Index ation
IP PATH Non Real Time Path Tx VS.IPPath.TxNrtBytes x

T.B.P.H1 [Mbps] T3W #N/A
Throughput 8/3600/10^6
IP PATH Non Real Time Path Rx VS.IPPath.RxNrtBytes x

T.B.P.H2 [Mbps] T3W #N/A
Throughput 8/3600/10^6
T.B.P.H3 IP PATH Non Real Time Path Tx pps(1) VS.IPPath.TxNrtPkts/3600 [pps] T3W #N/A
Site / T.B.P.H4 IP PATH Non Real Time Path Rx pps VS.IPPath.RxNrtPkts/3600 [pps] T3W #N/A
(Site &&
VS.IPPath.TxRtBytes x
IP_PATH) T.B.P.H5 IP PATH Real Time Path Tx Throughput [Mbps] T3W #N/A
8/3600/10^6
VS.IPPath.RxRtBytes x
T.B.P.H6 IP PATH Real Time Path Rx Throughput [Mbps] T3W #N/A
8/3600/10^6
T.B.P.H7 IP PATH Real Time Path Tx pps(1) VS.IPPath.TxRtPkts/3600 [pps] T3W #N/A
T.B.P.H8 IP PATH Real Time Path Rx pps VS.IPPath.RxRtPkts/3600 [pps] T3W #N/A
These KPIs have to be reported/calculated per site (as a general value) and per (SiteID && IP_PATH) individually
EE TEMPLATE FOOTER (1) pps = packets per second

11/09/2014 92
LTE Transport eNB side Layer 4 – capacity KPIs (cont.)

PRS
LTE Transport eNB side Layer 4 – capacity KPIs
Meas. Aggre
Object gation
T.B.C.L1 SCTP Rx Average Throughput Sctp.RxBytes x 8 /3600/1000 [Kbps] T3W #N/A
T.B.C.L2 SCTP Tx Average Throughput Sctp.TxBytes x 8 /3600 /1000 [Kbps] T3W #N/A
T.B.C.L3 SCTP Rx Maximum Throughput {Sctp.RxMaxSpeed} / 1000 [Kbps] T3W #N/A
T.B.C.L4 SCTP Tx Maximum Throughput {Sctp.TxMaxSpeed} / 1000 [Kbps] T3W #N/A
T.B.C.L5 SCTP Rx Minimum Throughput {Sctp.RxMinSpeed} / 1000 [Kbps] T3W #N/A

Site
T.B.C.L6 SCTP Tx Minimum Throughput {Sctp.TxMinSpeed} / 1000 [Kbps] T3W #N/A
100 x {Sctp.RxDropBytes} /
T.B.C.L7 SCTP Rx drop byte rate [%] T3W #N/A
{Sctp.RxBytes + Sctp.RxDropBytes}
100 x
T.B.C.L8 SCTP Rx packet drop rate [%] T3W #N/A
{Sctp.RxDropPkts}/{Sctp.RxPkts)
T.B.C.L9 SCTP Rx packets per second Sctp.RxPkts/3600 pps T3W #N/A
T.B.C.L10 SCTP Tx packets per second Sctp.TxPkts/3600 pps T3W #N/A


LTE Transport eNB side Layer 3 – IP routing
• A static host route is a manually configured route
whose subnet mask is 255.255.255.255.
• A static gateway route is a manually configured route

whose subnet mask is not 255.255.255.255 and whose
next hop IP address and egress port IP address are on
different network segments.
• A dynamic host route is a dynamically acquired route

whose subnet mask is 255.255.255.255.
• A dynamic gateway route is a dynamically acquired

route whose subnet mask is not 255.255.255.255 and
whose next hop IP address and egress port IP address
are on different network segments.

Meas. Aggreg
Obj. ation
T.B.I.L1 # IP Tx. packets VS.IP.TxPackets / 3600 [pps] T3W #N/A
If(VS.IP.TxPackets>10^5,VS.IP.TxDropPkts/V Wk.
T.B.I.L2 # IP Tx. packets discard rate
S.IP.TxPackets,0)
[%] Max
0.1%
T.B.I.L3 # IP Rx. packets VS.IP.RxPackets/10^6 [pps] T3W #N/A
Wk.
T.B.I.L4 # IP Rx. packets discard rate VS.IP.RxDropPkts/10^3 [%] Max
0.1%
eNB
T.B.I.L5 IP Tx. throughput VS.IP.TxBytes x 8 /10^6 / 3600 [Mbps] T3W #N/A
If(VS.IP.TxBytes>10^8, VS.IP.TxDropBytes / Wk.

T.B.I.L6 # IP Tx. Byte discard rate
VS.IP.TxBytes,0)
[%] Max
0.1%
T.B.I.L7 IP Rx. throughput VS.IP.RxBytes x 8 / 10^6 / 3600 [Mbps] T3W #N/A
If(VS.IP.RxBytes>10^8, VS.IP.RxDropBytes / Wk.

T.B.I.L8 # IP Rx. Byte discard rate
VS.IP.RxBytes,0)
[%] Max
0.1%
• If either of the Rx. KPIs in the table above in violation  it is possible that either the Radio I/F is
congested or that the (L/U)MPT or LBBP ports in congestion  please refer to the Radio I/F chapter
above for the appropriate corrective measures and KPIs.
• If either of the Tx. KPIs in the table above in violation  either the Transport segment is in congestion or
the eNB is actually in the Shaping region and the site is dropping  check the eNB Transport segment
KPIs (Layer 2) (defined next few slides).


Meas. Meas. Aggregati
KPI Index KPI name Formula on
SLAT
Object unit
T.B.M.H1 IP PM Average Tx Throughput {IPPM.Bits.MeansTx} / 10^6 [Mbps] T3W tbc
T.B.M.H2 IP PM Average Packet Tx Rate {IPPM.Pkts.MeansTx} / 10^3 / 3600 [Kpps](1) T3W tbc
T.B.M.H3 IP PM Max Tx Throughput {IPPM.Peak.Bits.RateTx} / 10^6 [Mbps] T3W tbc
T.B.M.H4 IP PM Max Packet Tx Rate {IPPM.Peak.Pkts.RateTx} / 10^3 /3600 [Kpps] T3W tbc
T.B.M.H5 IP PM Average Rx Throughput {IPPM.Peer.Bits.MeansRx} / 10^6 [Mbps] T3W tbc
T.B.M.H5 IP PM Average Packet Rx Rate {IPPM.Peer.Pkts.MeansRx} / 10^3 /3600 [Kpps](1) T3W tbc
T.B.M.H6 IP PM Max Rx Throughput {IPPM.Peer.Peak.Bits.RateRx} / 10^6 [Mbps] T3W tbc
T.B.M.H7 IP PM Max Packet Rx Rate {IPPM.Peer.Peak.Pkts.RateRx} 10^3 /3600 [Kpps] T3W tbc
eNB
IPPM(2) (([VS.IPPM.Local.TxPkts]- Wk.
T.B.M.H8 IP Forward packets loss rate [%] tbc
[VS.IPPM.Peer.RxPkts])/[VS.IPPM.Local.TxPkts])*{100} Max.
T.B.M.H9 IP Stdev forward jitter VS.IPPM.Forward.JitterStandardDeviation [ms] T3W tbc
T.B.M.H10 IP Stdev backward jitter VS.IPPM.Back.JitterStandardDeviation [ms] T3W tbc
T.B.M.H11 IP Average RTT VS.IPPM.Rtt.Means [ms] T3W tbc
T.B.M.H12 IP Maximum RTT VS.IPPM.MaxRttDelay [ms] T3W tbc
T.B.M.H13 IP Minimum RTT VS.IPPM.MinRttDelay [ms] T3W tbc
T.B.M.H14 IP Tx Packets VS.IPPM.Local.TxPKts/10^3 / 3600 [kpps] T3W tbc
T.B.M.H15 IP Rx Packets VS.IPPM.Peer.RxPkTs/10^3 /3600 [kpps] T3W tbc
(1) Kpps = Kilo Packets per Second (2) IPPM = IP Perf Mgmt – to be activated in Rel. 9
LTE Transport eNB side Layer 2 –
BT MEAS architecture and congestion points
2. VLAN
MSN PE1 • There is a direct Layer 2 connection between the eNB and
the MSN PEs  both eNBs and MSN PEs can be put in the
y.y.y.0/29
same subnet, VLAN respectively (Ethernet segment);
Physical I/F • All the direct connected routes from the Backhaul VLANs
VRRP H3G switch site
(y.y.y.8/29 and y.y.y.0/29) are being injected further on
into the AR routers (through OSPF);
VLAN MSN PE2 • The physical I/F IP address is also the VPN Tunnel end-
point address;
Detailed in • It is not possible to determine which BT MEAS port is

the H-QoS slide further on
active
MSN PE2
y.y.y.8/29 AR1
• BT ethernet transport service accepts ethernet frame
3. IDU ports sizes of 1522 bytes
Physical I/F
VRRP EE Switch site • The burst tolerances of the BT Etherflow service are
quoted as committed burst size (CBS) of 12 kbytes and
maximum burst size (MBS) of 60 kbytes i.e. these are the
VLAN AR2 burst sizes that can be tolerated without packet-loss in
1. ETHPORT MSN PE2 the BT Ethernet service.
y.y.y.0
• /29 mask  network mask is 255.255.255.248  valid subnets (VLANs) are: y.y.y.8
 within each subnet max. 6 hosts; y.y.y.16
EE TEMPLATE FOOTER 11/09/2014 ………. 100
LTE Transport eNB side Layer 2 –
VM architecture and congestion points
2. VLAN
MSN PE1
• This is a Layer 3 solution, so there is direct IP
x.x.x.0/30
BFD “visibility” between the eNB and the MSN PEs
z.z.z.0/30
Physical I/F CSG  they can be put in the same subnet
H3G switch site (Ethernet segment);
VLAN MSN PE2 • All the direct static routes from the Backhaul
x.x.x.4/30 VLANs (z.z.z.4/30 and z.z.z.0/30) are being
injected further by the MSN PE routers onto
Detailed in
the H-QoS slide further on the AR routers (through OSPF);
MSN PE2
x.x.x.8/30
AR1 • The physical I/F IP address is also the VPN
z.z.z.4/30 BFD Tunnel end-point address;
CSG
Physical I/F
EE Switch site • BFD used on MSN PE1 to ensure VLAN
availability;
VLAN AR2
1. ETHPORT x.x.x.12/30 MSN PE2
z.z.z.0
• /30 mask  network mask is 255.255.255.252  valid subnets (VLANs) are: z.z.z.4
 within each subnet max. 2 hosts; z.z.z.8
EE TEMPLATE FOOTER 11/09/2014 ………. 101
LTE Transport eNB side Layer 2 – eNodeB shaping
Dimensioning upstream
• Two types of eNB sites: link for site k
• Aggregation (non-dependency) Downstream
sites
• Microwave (dependency);
• The eNB QoS settings and the dimensioning is different,

function of site type;
eNBk
• Aggregation sites: the eNB QoS shaping is presented in the

table below;
Total bwidth
• Microwave sites: the main condition to be fulfilled per site:
𝑆ℎ𝑎𝑝𝑖𝑛𝑔_𝑅𝑎𝑡𝑒𝑘 ≤ 𝑇𝑜𝑡𝑎𝑙𝑏𝑤𝑖𝑑𝑡ℎ − Σ(𝑆ℎ𝑎𝑝𝑖𝑛𝑔_Ratei), where i=1..(k-1)
The shaping rate for a site should not exceed the total upstream backhaul
capacity take away the sum of shaping rates for all the sites which are
downstream from the site in discussion (as presented in the right figure).
• Where eNodeBs are connected to the Hub site using

microwave technology all eNodeBs (including Hub site
one) reside on the same VLAN

Meas. Aggrega
Obj. tion
T.B.E.L1 FE Average Tx Throughput {FEGE.TxBytes} x 8 / 10^6 / 3600 [Mbps] T3W A
T.B.E.L2 FE Average Rx Throughput {FEGE.RxBytes} x 8 / 10^6 / 3600 [Mbps] T3W #N/A
T.B.E.L3 FE Average Tx Packets Rate {FEGE.TxPackets} / 10^3/3600 [Kpps] T3W #N/A
T.B.E.L4 FE Average Rx Packets Rate {FEGE.TxPackets} / 10^3/3600 [Kpps] T3W #N/A
T.B.E.L5 Fast Ethernet Max Tx Throughput {FEGE.TxMaxSpeed} / 10^6 [Mbps] T3W 80
eNB T.B.E.L6 Fast Ethernet Max Rx Throughput {FEGE.RxMaxSpeed} / 10^6 [Mbps] T3W 80
T.B.E.L7 Fast Ethernet Min Tx Throughput {FEGE.TxMinSpeed} / 10^3 [kbps] T3W #N/A
T.B.E.L8 Fast Ethernet Min Rx Throughput {FEGE.RxMinSpeed} / 10^3 [kbps] T3W #N/A
T.B.E.L9 FE Total Tx bandwidth {FEGE.TxTotalBW}/10^3 [Mbps] #N/A #N/A
T.B.E.L10 FE Total Rx bandwidth {FEGE.RxTotalBW}/10^3 [Mbps] #N/A #N/A

100X[{FEGE.RxErrPackets} /
T.B.E.L11 Fast Ethernet Rx. Error Packet Rate
{FEGE.RxPackets+FEGE.RxErrPackets}]
[%] Wk. Max 0.1%
• If T.B.E.L5 in violation  consider shaping activation for (eNB  Core) direction;

• If T.B.E.L6 or T.B.E.L11 in violation  consider tweaking the H-QoS in CX600;
• If neither of the two solutions above possible, consider the Transport link expansion.


Annexes

Annex 1.
More details on the architecture

Huawei recommendations: The LTE RAN congestion points (cont.):
Transport congestion points
TCSW Control Plane User Plane 1. S1 link eNodeB
S1-AP
PDCP APPLIC GTP-U
4. X2 link X2-AP 2. GTP-U
tunnel S1 UP
TRANSP SCTP UDP 3. IP PATH SGW
S1/X2 CP S1/X2 eNode

IP adr
5. SCTP
Original IP UP IP addr X2 UP B
NETW association
ESP
SCTP assoc IP Path
(S1/X2 CP) Transport IP (S1/X2 UP)
8. VLAN MME
VLAN
DATA
Transport LLC eNode
IP addr
Q P B
LINK
C H MAC
I B 6. ETHPORT
PHY ETH
FE/GE0 750Mbps
Ipsec Tunnel end-point
Sec-GW
IP address
Air I/F Transport Common7. IPSecCommon
IPSec IPSec
Tunnel TRG Tunnel
shaper
References
[1] - LTE – The UMTS Long Term Evolution From Theory to Practice, Second Edition - Stefania
Sesia; Issam Toufik - John Wiley & Sons, September 6, 2011, Print ISBN 978-0-470-66025-6

Change history
[1] – 14/Aug/2014  CAPS limitation KPIs added to the site KPIs

THANK YOU

Lte Capacity Management v0.1 MBNL Esl

Uploaded by

Copyright:

Available Formats

Lte Capacity Management v0.1 MBNL Esl

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lte Capacity Management v0.1 MBNL Esl

Uploaded by

Copyright:

Available Formats

LTE CAPACITY MANAGEMENT

KPIs and Dimensioning rules

11/09/2014 Remus Sepp

EE TEMPLATE FOOTER 11/09/2014 2

EE TEMPLATE FOOTER 11/09/2014 3

EE TEMPLATE FOOTER 11/09/2014 5

EE TEMPLATE FOOTER 11/09/2014 8

EE TEMPLATE FOOTER 11/09/2014 9

Web III. Infrastructure domain (PKI and DHCP servers)

Radio I/F Transport

Site Cell Backhaul Connectivity Infrastructure

EE TEMPLATE FOOTER 11/09/2014 11

EE TEMPLATE FOOTER 11/09/2014 12

EE TEMPLATE FOOTER 11/09/2014 13

Transport LMPT LBBP

c3. SRS Resource

EE TEMPLATE FOOTER 11/09/2014 15

In-sync RRC Out-of-sync 800/cell, 800/cell, 800/cell,

Out-of-sync DL/UL max 225/100 450/225

MIMO 2x2 2x2 4x4, 4rxdiv

1xLBBPc 3 1 300 100

2xLBBPc 6 2 73/150 750 43/70 750

EE TEMPLATE FOOTER 11/09/2014 18

Processing Expand the

(1)  this is a “per-eNB” figure

EE TEMPLATE FOOTER 11/09/2014 19

LMPT/UMPT module Σ(L.Traffic.User.Max), per (L/U)MPT Weekly LMPT 1080

EE TEMPLATE FOOTER 11/09/2014 22

LMPT 30CAPs 30CAPs 60CAPs

EE TEMPLATE FOOTER 11/09/2014 23

Average LBBP-c Σ(CAPS_Formula)/3600, across all cells Call atts

Primary KPI (1)  T3W = Top 3 Weekly value,

LMPT/UMPT module LMPT 1080

LMPT/UMPT module total LMPT 270

Percentage of times that

LBBP CPU load (on each

EE TEMPLATE FOOTER 11/09/2014 26

EE TEMPLATE FOOTER 11/09/2014 27

EE TEMPLATE FOOTER 11/09/2014 28

• If a UL unsynchronized UE requests uplink

EE TEMPLATE FOOTER 11/09/2014 29

Failed >70% and

Failed >60% and Set PucchSwitch:ON

(1) – please check the next slide for the PRACH

Set PDCCH Symbol

User plane and

EE TEMPLATE FOOTER 11/09/2014 31

#users per cell L.Traffic.User Avg./Max. x x Active

# users in UL sync L.Traffic.User x x Ulsync.Avg Active

EE TEMPLATE FOOTER 11/09/2014 32

EE TEMPLATE FOOTER 11/09/2014 33

Avg DL User throughput L.Thrp.bits.DL / 10^6

Avg. DL signalling L.Thrp.bits.UL.SRB /

PCH Throughput {L.Traffic.PCH.TB.bits} / 3600 /

EE TEMPLATE FOOTER 11/09/2014 35

R.C.M.L1 QCI1 DL Mbps L.Thrp.bits.DL.PLMN.QCI.1(bit) / 10^6 / 3600 [Mbps] T3W tbc

R.C.M.L2 QCI2 DL Mbps L.Thrp.bits.DL.PLMN.QCI.2(bit) / 10^6 / 3600 [Mbps] T3W tbc

R.C.M.L3 QCI3 DL Mbps L.Thrp.bits.DL.PLMN.QCI.3(bit) / 10^6 / 3600 [Mbps] T3W tbc