Short Paper—4.

5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

4.5G Capacity & QoS Optimization: A New Proposal to

Enhance the Principal KPIs of 5G/IoT
https://doi.org/10.3991/ijoe.v14i11.9256
Ahmed El Mahjoubi(), Tomader Mazri, Nabil Hmina

National School of Applied Sciences Kenitra, Morocco

[email protected]

Abstract—RF capacity management and optimization is an important part of the

network performances of the future. Technically, 4.5G/5G mobile network is
expected to enable people-thing and thing-thing interconnections by combining
communication technologies and networks. Our Lab has been actively promoting
research on 4G & 5G performances, in which the RF Capacity management issue is
of great interest. New 4G/5G tests put forward a lot of requirements for RF network
performances in terms of KPIs, QoS, as well as user experience. Based on
Morocco’s current situation, this article first discusses the LTE cell resources
monitoring and optimization, Furthermore, we will introduce LTE OMSTAR tool,
and LTE cell resources optimization scenarios are summarized.

Keywords— LTE Capacity, RF Optimization, 5G, QoS, IoT.

1 Introduction

With the rapid development of smart terminals and services the impact of capacity
problems on KPIs and User experience increased, therefore close monitoring for Key
bottleneck of LTE resources is required to ensure better KPIs and user experience.
Capacity monitoring can be implemented using the following two methods: [1]

• Daily monitoring for prediction: Counters are used to indicate the load or usage of
various types of resources on the LTE network. Thresholds for resource consump-
tion are specified so that preventive measures such as reconfiguration and
expansion can be taken to prevent network congestion when the consumption of a
type of re-source continually exceeds the threshold. [2]
• Problem-driven analysis: This method helps identify whether a problem indicated
by counters is caused by network congestion through in-depth analysis. With this
method, problems can be precisely located so that users can work out a proper net-
work optimization and expansion solution.
LTE Resources are:

iJOE ‒ Vol. 14, No. 11, 2018 193

Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

Fig. 1. LTE cell and eNodeB resources.

This paper concentrates on LTE Cell Resources.

2 PRB: Physical Resource Block

Resource Block is the minimum unit for resource allocation used for data transmis-
sion in physical layer, it consists of two dimensional domain: [3]
• Time Domain: 1 Slot= 7 OFDM Symbol (normal CP).
• Frequency Domain: 12 Subcarriers.
RB number depend on the LTE system bandwidth as shown in table 1

Table 1. RB number vs LTE system bandwidth.

Channel bandwidth BWChannel [MHz] 1 3 5 10 15 20

Sub-carriers Number 72 180 300 600 900 1200
RB Number 6 15 25 50 75 100

2.1 Resource Element (RE)

Smallest unit in PRB consist of two dimensional domain:

• Time Domain: 1 OFDM Symbol.

• Frequency Domain: 1 Subcarriers.

194 http://www.i-joe.org
Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

Fig. 2. LTE resource element.

2.2 Usage Increase by

• As Data traffic growth PRB Usage increases.

• As PRB Usage reaches 100%, user perceived traffic will decrease.

2.3 Insufficiency

• Growing traffic leads to a continuous increase in PRB usage. When the PRB usage
approaches to 100%, user-perceived rates will decrease as follow:
─ New Users may fail to be admitted
─ Admitted users experience is affected.

2.4 Monitoring

• Downlink PRB usage = L.ChMeas.PRB.DL.Used.Avg/L.ChMeas.PRB.DL.Avail x

100%
─ L.ChMeas.PRB.DL.Used.Avg = Average used DL PRBs
─ L.ChMeas.PRB.DL.Avail = Available DL PRBs
• Downlink user-perceived rate (Mbit/s) = L.Thrp.bits.DL/L.Thrp.Time.DL/1000
─ L.Thrp.bits.DL = DL PDCP Layer Throughput

iJOE ‒ Vol. 14, No. 11, 2018 195

Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

─ L.Thrp.Time.DL= duration for transmitting DL PDCP layer data

2.5 Root Cause Analysis/Solution

• Factors affecting PRB usage:

─ Poor Channel Quality
─ Parameter Settings
─ User number
─ Traffic Model ”service requirement of each user”
• PRB Limitation leads to prolong UE Scheduling delay  decreasing UE Data Rate.

2.6 Poor Channel Quality

• Lead to low spectrum efficiency and increase PRB usage.

• DL Spectrum Efficiency = L.Thrp.bits.DL/L.ChMeas.PRB.DL.Used.Avg/ (rep rt
Period*60)/180000 “bit/s/Hz”
• If Less than 1bit/s/Hz  Optimize RF Quality
• UL Spectrum Efficiency = L.Thrp.bits.UL/L.ChMeas.PRB.UL.Used.Avg/ (report
Period*60)/180000 “bit/s/Hz”
• If Less than 0.5bit/s/Hz  Optimize RF Quality

2.7 Parameter Settings

• DL
─ RbgAllocStrategy=Adaptive
─ If the number of required RBs is less than that of one RBG, RBs are allocated to
the scheduled UE as required, which is specified by resource allocation type 1.
─ If the number of required RBs is greater than that of one RBG, the number of
allocated RBGs is rounded up and an integral number of RBGs are allocated to
the scheduled UE. Compared with RBG round-up, this mode prevents RB waste
when the number of required RBs is less than that of one RBG
• UL
─ UlSchSwitch=PuschDtxSwitch-1
─ During UL transmission if UE fail to detect UL GRANT or SR false alarm
appear in DRX state
─ Enabled:UE will report(ACK/NACK/DTX) if the HARQ feedback state of one
uplink schedule is DTX, eNodeB will deliver UL GRANT again instead of start-
ing HARQ retransmission. If the next HARQ feedback is still DTX, eNodeB
will stop the schedule triggered by this SR
─ Disabled:lead to the uplink retransmission repeatedly and waste the uplink
PUSCH resource and the uplink error code rate will increase obviously.

196 http://www.i-joe.org
Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

2.8 User Number

• If average RRC_Connected User(L.Traffic.User.Avg) >300

─ Downlink user-perceived rate < a user-defined threshold (default value: 2 Mbit/s)
• Resource Insufficiency cause the low Data Rate Consider Expansion
• Formula for Downlink user perceived rate:
─ Average DL Throughput for single User =
(L.Thrp.bits.DLL.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI*1000

2.9 Threshold for Expansion

• Add carriers or eNodeBs if both of the following conditions are met:

─ Downlink PRB usage ≥ 70%
─ Downlink user-perceived rate < a user-defined threshold (default value: 2 Mbit/s)

2.10 OMSTAR Analysis: Operation & Maintenance

• Whole Network PRB Analysis

• PRB Resource Usage Distribution
• Uplink/Downlink PRB Resource Usage Distribution (RRU)
• The Cell List of Uplink/Downlink PRB Resource Usage Over Threshold
• PRB Resource Usage(Daily)(Cell/Busy Hour)(Network/Busy Hour)

2.11 UL / DL Formulas

• Uplink PRB resource usage = Average number of uplink PRBs scheduled in each
second
• (L.ChMeas.PRB.UL.Used.Avg+ L.ChMeas.PRB.PUCCH.Avg)/ Total number of
uplink PRBs in a cell x 100%
• Downlink PRB resource usage = Average number of downlink PRBs scheduled in
each second
• (L.ChMeas.PRB.DL.Used.Avg)/ Total number of downlink PRBs in a cell x 100%

2.12 OMSTAR Threshold is

• OMSTAR threshold for Expansion is 70% PRB usage.

Table 2. PRB cell usage KPIs.

Whole Network Whole Network Whole Network Whole Network

Average Number Average Number Average Uplink Average Downlink
Time
of Uplink Used PRB of Downlink Used PRB Resource PRB Resource
per Cell PRB per Cell Usage per Cell Usage per Cell
x 4.33 4.85 8.65% 9.71%

iJOE ‒ Vol. 14, No. 11, 2018 197

Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

2.13 For Low Spectral Efficiency

DL (1bPs/Hz):
Check if MCS <10 or CQI<6  RF Issues  otherwise Check MIMO
Setting UL (0.5 bps/HZ):
Check if MCS<10 or SINR<5  RF Issues
Downlink/Uplink Spectral Efficiency (Daily) (Cell/BH) (Network/BH)

3 PRACH: Physical Random Access CHannel

Random access is performed in the following scenarios: [3]

1. Initial RRC connection setup to switch from the RRC_IDLE state to the
RRC_CONNECTED state, a UE initiates random access.
2. RRC connection reestablishment When a radio link failure (RLF) occurs, the UE
needs to reestablish an RRC connection. In this scenario, the UE initiates random
access.
3. Handover During a handover, a UE initiates random access in the target cell.
4. Downlink data arrival when an eNodeB needs to send downlink data to a UE in the
RRC_CONNECTED state and finds that the UE is out of uplink synchronization,
the eNodeB instructs the UE to initiate random access.
5. Uplink data arrival When a UE in the RRC_CONNECTED state needs to send up-
link data to an eNodeB and finds that it is out of uplink synchronization, the UE
initiates random access.
6. LCS When a location service (LCS) is required, the eNodeB initiates random access.

3.1 Random Access message

As shown in below figure consist of

• CP ”Cyclic prefix”
• preamble sequence
• Guard period
Since Random Access message is first message sent from UE side and the TA
“Time advance” is not known yet, so guard period is used to ensure message arrival is
within the TS.

CP Sequence

T T
CP SEQ Guard Period
Preamble

Fig. 3. Random Access message

198 http://www.i-joe.org
Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

3.2 Usage Increase by

• Access requests (initial- UL/DL Data).

• Handover times.
• LCS
• RRC connection reestablishment.

3.3 Root Cause analysis/Solution

Enable Random Access Backoff. The RACH does not interfere with other uplink
channels in LTE. Generally, the probability of RACH conflicts is low.
If excessive UEs are admitted on a PRACH, however, preamble conflicts may oc-cur, and
some UEs may fail to access the network. To solve this problem, random ac-cess backoff
is introduced to control the time for preamble retransmission by UEs. [4] If random access
backoff is enabled, the eNodeB notifies a UE of a backoff value in the random access
response message. When the UE needs to retransmit a preamble, it randomly selects a
value (from 0 to the received backoff value) as its backoff time. The
UE can retransmit the preamble only after the backoff time ends.
Random access backoff is not performed in the following two scenarios:

• Initial preamble transmission

• Preamble retransmission in non-contention-based random
access When to use?
If the random preamble usage reaches or exceeds 75% for three days by default in
a week, enable the adaptive backoff function.
PRACH resource adjustment algorithm. In case of Preambles are insufficient
PRACH resource adjustment algorithm acts as follow:
If dedicated preambles are in surplus and random preambles are insufficient, the
eNodeB reduces the number of dedicated preambles.
If dedicated preambles are insufficient, the eNodeB increases the number of dedi-
cated preambles.
Whether dedicated preambles are insufficient is measured based on the dedicated
preamble allocation failure rate, which is equal to one minus the ratio of the number
of UEs that are allocated dedicated preambles to the number of UEs that apply for
dedi-cated preambles.
MaksIdxSwitch: Indicates the switch used to control reuse of dedicated preambles
between UEs. If the switch is on, the eNodeB enables reuse of dedicated preambles
among UEs based on the MaskIndex parameter. If the switch is off, the eNodeB allo-
cates a dedicated preamble to only one UE at a time.
When to use?
If the dedicated preamble usage reaches or exceeds 75% for X days (three days by
default) in a week, enable the PRACH resource adjustment algorithm and reuse of
ded-icated preambles between UEs.

iJOE ‒ Vol. 14, No. 11, 2018 199

Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

This helps reduce the probability of UEs initiating contention-based random access
in the case of dedicated preamble insufficiency and therefore helps reduce the access
delay.
Threshold for Expansion. If above actions implemented and the resource usage is
still >75% add eNodeB or split Cell.

4 Profit from LTE RF Capacity Management and Optimization

By monitoring, managing and optimizing LTE Capacity you can:

• Avoid Access Failure

• Improve UE Experience and main 4G/4.5G/5G KPIs
• Decrease Handover Delay
• Decrease Uplink and downlink scheduling delays.

5 Conclusion

This paper provides an overview of LTE capacity optimization steps and provide a
new approach for 4.5G/5G three main KPIs improvement (massive connections, low
latency and high throughput).This article discusses also RF optimization solutions of
LTE cell resource. Our analysis results by OMSTAR Tool show that the targets can
be achieved in all LTE expansion scenarios to 4.5G and 5G/IoT.

6 Acknowledgment

This paper is supported by Huawei and Ericsson, 5G/IoT Tech Lab, Maroc
Telecom R&D Center.

7 References
[1] 4.5G, Opening Giga Mobile World, Empowering Vertical Markets, Huawei, white paper,
2015.
[2] ICT-317669 METIS project, Updated scenarios, requirements and KPIs for 5G mobile and
wireless system with recommendations for future investigations, Deliverable D1.5, April
2015.
[3] G. Gomez et al., “Towards a QoE-Driven Resource Control in LTE and LTE-A
Networks,” J. Computer Networks and Commun., 2013.
[4] R1-157647, “Summary of evaluation results”, Huawei, HiSilicon,

200 http://www.i-joe.org
Short Paper—4.5G Capacity & QoS Optimization: A New Proposal to Enhance the Principal KPIs …

8 Authors

Dr. Ahmed EL MAHJOUBI received state engineer degree in networks & tele-
communication systems from the National School of Applied Sciences Kenitra, Mo-rocco,
in 2014. Since 2014 he has been a Technical Director in the Department of Net-work
Performances Service (NPS), Huawei Technologies. His main research interests include
3G/4G/5G systems QoS/QoE/KPIs, Network Audit, Network Planning and Op-timization,
wireless communication systems especially in the topics of RF systems.
He has been involved in several national and international MBB projects on RF Sys-
tems: IAM 2G/3G SingleRAN Project, IAM LTE & VoLTE Project, IAM GUL Pro-ject,
INWI 3G/4G Project, INWI UMTS900 Project and Bouygues Telecom 4G/5G Project. He
has been the radio coordinator of the first test of 4G+ in Morocco & Africa.
Prof. Tomader Mazri, HDR degree in Networks and Telecommunication from
IbnTofail University, Ph.D. degree in Microelectronics and Telecommunication from
SidiMohamed BenAbdellah University and INPT of Rabat, Master’s degree in Micro-
electronics and Telecommunication Systems, Bachelor’s degree in telecommunication
from CadiAyyad university. She is currently a professor at National School of Applied
Sciences of Kenitra and a Permanent member of Electrical and Telecommunications
Engineering Laboratory. Author and co-author of twenty articles journals, forty articles in
international conferences, a chapter and three books. Her major research interests are on
Microwave systems for mobile and radar, Smart antennas and NG Mobile network.
Prof Nabil HMINA, Professor of Higher Education Director of the National
School of Applied Sciences Kenitra, since November 2011 to date, Degree in Physics,
Option: Thermodynamics - Mohammed V University, University PhD - Engineering
Sciences, University and Ecole Centrale de Nantes, 1994, HDR (1st in Morocco) Ibn
Tofail Uni-versity, Kenitra, 2002. Vice-President for Academic Affairs and
Information Technol-ogy of the University Ibn Tofail, 2005-2011 Post-Doctoral
Researcher: EDF - thermo-kinetics Laboratory of Nantes, 1994-1995, Research
Engineer at PolyTech school of Nantes, 1995-1997. Director of the research
laboratory "Systems Engineering" since 2012 (15 permanent teachers 10 teachers
Researchers Associate researchers 75 PhD students.
Author and/or co-author of a dozen articles in notorious newspapers and forty
papers in international conferences. Jury President, member of several theses and
Habilitations for research orientation.

Article submitted 21 July 2018. Resubmitted 20 October 2017. Final acceptance 21 October 2018. Final
version published as submitted by ther authors.

iJOE ‒ Vol. 14, No. 11, 2018 201