a tutorial on SON when applied to small cells as they are expected to be widely deployed in 5G networks.

1. Self Organising Network (SON) Techniques for Small
Cells in 5G
Karim Nasr & Klaus Moessner

2. Overview
Thursday, 24 August 2017 2
• 5G Vision and Key Enabling Technologies
• Small Cells and deployment challenges
• SON techniques and features
• Case Study : MLB
• Shared Spectrum Issues
• Challenges and Open Issues
• Summary

3. 5G Vision
Thursday, 24 August 2017 3
• Higher data rates (capacity and throughput improvement).
• Improved spectrum efficiency (bps/Hz/m2).
• Enhanced end user Quality of Experience (QoE) with a wide variety of
requirements including traditional QoS requirements, reliability, security and
others.
• Reduced end to end latency.
• Seamless and improved coverage and mobility.
• Lower energy consumption / improved energy efficiency (Green radios).

4. 5G Key Enabling Technologies
Thursday, 24 August 2017 4
Expectations and Features Enabling Technologies and Trends
Capacity and throughput improvement, high data rate
(~1000x of throughput improvement over 4G, cell data rate
~10 Gbps, reduced signalling overheads)
Spectrum reuse and multiband/multi technology operation (from UHF
to mm-wave/visible light communication bands),
Heterogeneous / multi-tier networks , small cells
Multi RAT RRM/ C-RAN , SON
Massive-MIMO, new air interface for spectral efficiency
Reduced latency
(2~5 milliseconds end-to-end latencies)
Multi RAT RRM/ C-RAN , SON, D2D communication, Full-duplex
communication
Network densification: Ultra Dense Networks
(~1000x higher mobile data per unit area, 100~10000x higher
number of connecting devices/users)
Heterogeneous / multi-tier networks , small cells
Multi RAT RRM/ C-RAN, SON, Seamless operation
Advanced services and applications
(e.g., smart city, service-oriented communication, IoT)
Multi RAT RRM/ C-RAN, SON
network virtualization, M2M communication
Improved energy efficiency /green radios
(~10x prolonged battery life)
Wireless charging, energy harvesting
Multi RAT RRM/ C-RAN , SON
Autonomous applications and network management,
Internet of Things
SON/ cognitive networks
M2M/ D2D communication

5. Dense Small Cells
Thursday, 24 August 2017 5
v Low Power nodes but high quality in terms of propagation channel conditions (much closer to
mobile users at the edge of coverage resulting in a better throughput)
v Co-channel Deployment (needs intelligent RRM techniques)
v Densification: An effective approach to high capacity provision under limited spectrum
resources is to densely deploy small cellular base stations
v Contribute to the 5G capacity targets
Macro cell
Small cells

6. Dense Small Cells
Thursday, 24 August 2017 6
• The deployment of small cells (pico and femto cells) is usually done in a decentralised plug and play fashion.
• Reduces the need for busy hour capacity in the macro network layer.
• Improves indoor and outdoor coverage and reduces overall network power consumption (improves energy efficiency)
• Provides several folds capacity increase is the areas of high demand, and reduces the service provider overall network
CAPEX and OPEX costs.

7. Dense Small Cells: Market Status
Thursday, 24 August 2017 7
Source: Small Cell Forum, Market status report, 2016

8. Main Challenges for Small Cell Deployment
Thursday, 24 August 2017 8
Network Complexity Management
• Traffic load balancing problems between macro cell and the small cell tiers or among small cells.
• Mobility management.
• Backhaul congestion management issues.
• Self Organising Network (SON) Techniques
Spectrum and Radio Access Management
• Physical and medium access control layers issues such as
• Co-tier and cross-tier interference mitigation
• Intelligent Radio Resource Management (RRM) / SON Techniques
• Cognitive Radio / Dynamic Spectrum Access (DSA)
• Delivering reliable QoS as well as reducing signalling overhead in a dynamic radio
network/environment
Energy Efficiency
• Intelligently control the number of activated cells based on the dynamics of user traffic, as well
as maintaining adequate QoS and capacity

9. SESAME: Key technologies and scenarios
Thursday, 24 August 2017 9
v An H2020 project on small cells
v Small cEllS coordinAtion for Multitenancy and Edge services
v Small Cells as a Service (SCaaS)
v Key technologies:
o Network Function Virtualization (NFV): software implementation of network functions
at the network edge bringing required flexibility and allowing multiple tenants.
o SON : Self X functions
o Mobile Edge Computing (MEC) / light DC
o New sharing models (VSCNO sharing same infrastructure and CESC)
v Use cases and scenarios examples:
o Large business centres
o Mobile end user generating HD real time content
o Sudden high concentration hot spots (e.g. stadium, conference centre, exhibition,
carnival, … etc.)

10. Self Organising Networks (SON): The Need
Thursday, 24 August 2017 10
• Very large number of small cells to be deployed.
• Manual processes for configuration and optimisation are no longer practical/feasible
• Dynamic deployment requires quick and frequent adaptation
• Continuous adjustments of parameters during operation based on actual measurements and KPIs
Main Targets:
• Keeping operational effort at an acceptable level/ ideally eliminate user intervention.
• Protecting network operation by reducing the probability of errors
• Speeding up the planning, configuration, management, optimisation and healing of mobile
communications networks.
[1] T Q. S. Quek, G. de la Roche, İ. Güvenç, M. Kountouris, “Small Cell Networks Deployment, PHY Techniques, and Resource
Management” , Cambridge University Press, 2013
[2] A. J. Fehske, I. Viering,J. Voigt, C. Sartori, S. Redana and G.P Fettweis, ”Small-Cell Self-Organizing Wireless Networks,”
Proceedings of the IEEE, vol. 102, no. 3, pp.334-350, March 2014.
[3] A. Anpalagan, M. Bennis, R. Vnnithamby, “Design and Deployment of Small Cell Networks”, Cambridge University Press, 2016

11. SON Main Architectures
Thursday, 24 August 2017 11
• Distributed SON (D-SON)
• Small scale short term techniques
• Reacts to problems such as handover failures (time scales of seconds)
• Performed locally at BS’s based on information exchanged between neighbours
• Centralised SON (C-SON)
• Large scale and longer term techniques
• Jointly adjust parameters of an entire cluster of cells to daily traffic variations
• Requires some central coordination to improve overall network capacity based
on long term average values
• Hybrid SON (H-SON)
• A mix of D-SON and C-SON

12. SON: Self X Functions
Thursday, 24 August 2017 12
Source: Seppo Hämäläinen, Henning Sanneck, Cinzia Sartori, “LTE self-organising networks (SON) : network management
automation for operational efficiency” John Wiley , 2012
Self
Healing
Self
Optimisation
Self
Configuration/Planning
Alarm correlation
Root cause analysis
Sleeping-cell detection
Cell outage compensation
Coverage and capacity optimisation
Inter-cell interference coordination
Energy saving
Auto connectivity/configuration
Dynamic radio configuration
Automatic neighbour cell configuration

13. Self Configuration/Planning
Thursday, 24 August 2017 13
• Process of bringing a new cell into service with minimal human operator intervention.
• Plug and Play based operation
• Three main phases:
• Auto connectivity
• Auto commissioning
• Dynamic radio configuration (Physical Cell ID, Automatic neighbour relation (ANR)
for handover, initial power and antenna tilt / azimuth settings)
• New cells are automatically configured and integrated into the network
• Connectivity establishment and download of configuration parameters are software
based.
• When a new cell is introduced into the network and powered on, it gets immediately
recognised and registered by the network.
• The neighbouring cells automatically adjust their technical parameters (such as
emission power, antenna tilt, etc.) in order to provide the required coverage and
capacity, and, in the same time, avoid the interference.

14. Self Optimisation
Thursday, 24 August 2017 14
• Further optimisation steps are necessary during the operation of the network due to the fact that the
environment may change as a result of :
• Propagation conditions (e.g. new buildings, changes due to atmospheric conditions, vehicles).
• Traffic behaviour (e.g. new traffic concentrations).
• Deployment (e.g. the insertion of new cells).
• Previously configured parameters will become suboptimal.
• Adaptation of the parameters to track changes can improve the performance of the network.
• Practical examples:
• Automatic switch-off of a percent of cells during night hours will change ANR tables. Cells would
then re-configure their parameters in order to keep the entire area covered by the signal.
• In case of a sudden growth in connectivity demand (conference , stadium), the "sleeping" base
stations "wake up" almost instantaneously.
• Energy savings implications

15. Examples of Self Optimisation Features
Thursday, 24 August 2017 15
• Mobility Robustness Optimisation (MRO) guarantees proper mobility for users, i.e. proper
handovers and re-selection between cells of the same, but also of a different RAT.
• Mobility Load Balancing (MLB) and Traffic Steering try to optimally distribute traffic over cells
due to load condition, but also due to other properties such as speed, QoS or
energy consumption.
• Energy Saving Management is achieved on both network and UE side, for example, through switching
off inactive network nodes or reducing transmit power.
• Coverage and Capacity Optimisation (CCA) continuously adapts in particular antenna tilts and
transmit powers to maximise coverage, but also to optimise capacity through minimising
interference between the cells.
• RACH Optimisation: The Random Access Channel (RACH) needs to be accurately configured to
provide sufficient number of random access opportunities to UEs in any of the possible cells. The aim is
to find the best trade-off between performance and the resources which have to be sacrificed.

16. Self Healing
Thursday, 24 August 2017 16
• If one of the network elements/cells is faulty, there will be no other entity to offer
service until the fault is rectified.
• Faults can be due hardware, software, network planning and configuration errors or
due to environmental factors.
• During the resulting period of degraded performance, users are not experiencing
services with acceptable availability, reliability or quality-of-service (QoS), which may
cause serious revenue loss for the operator.
• Self healing aims at reducing the impacts from the failure, for example by adjusting
parameters and algorithms in adjacent cells so that other nodes can support the users
that were supported by the failing node.

17. Examples of Self Healing Features
Thursday, 24 August 2017 17
• Self recovery of software faults: fall back to a previous working version or initial configuration.
• Self recovery of hardware faults: can be achieved if a backup board can take over after a reset.
• Cell outage detection: Several system variables, performance indicators, alarms are continuously
monitored and compared against thresholds and profiles. This enables the detection of sleeping
cells/ cells out of service automatically.
• Cell outage recovery: The system recovers a cell outage automatically. Based on detection and
diagnosis result the best available recovery action (e.g. a cell reset) is performed and the operator
is notified about the results.
• Cell Outage Compensation : defines the system’s ability to compensate a cell outage
automatically to maintain as much as possible normal services to subscribers. First the actual
situation is studied by collection of the available configuration information. Then the associated
cells are reconfigured to improve service quality in the coverage area of the cell in outage.

18. Case Study : Mobility Load Balancing (MLB) in Small Cells
Thursday, 24 August 2017 18
The problem: Uneven Load Distribution !
0
50
100
1 2 3 4 5 6 7
Load %
Cell Number

19. Mobility Load Balancing (MLB)
Thursday, 24 August 2017 19
v MLB is a SON (Self-Organising Network) algorithm: A self optimisation
functionality
v Addresses the problem of uneven traffic or load distribution.
v Objective is to intelligently spread user traffic across systems radio
resources, to ensure QoS, by reducing call blocking and improving edge-
user throughput
v Enables overloaded cells to re-direct a percentage of their load to
neighbouring less loaded cells hence alleviating congestion problems.

20. Mobility Load Balancing (MLB) Approaches
Thursday, 24 August 2017 20
v Ideally, participating cells have different usage patterns with respect to time.
v Depending on the network scenario, MLB can postpone the deployment of
additional network capacity hence reducing costs (CAPEX).
v Standard MLB makes use of the Cell Range Expansion (CRE)
– CRE can be achieved by either cell coverage parameter adjustments or mobility (HO)
parameter adjustments
– Can provide real time optimization of cell overload through HO of cell-edge UEs (in
Idle/connected modes) to cell(s) with spare capacity
v Advanced MLB makes use of CRE together with the Almost blank Subframes
(ABS) feature

21. On distributed cell-association traditional schemes
Thursday, 24 August 2017 21
CRE scheme:
● Increases the downlink coverage footprint of low-power BSs (biased BSs) by adding a
positive bias
● Off-loaded users may experience unfavourable channel from biased BSs and strong
interference from unbiased high-power BSs
● Trade-off between cell load balancing and system throughput depends on the bias values
(typically 6 to 9 dB)
● CRE just forces alternate cell selection – It does not consider resource allocation (nor
loading, distance, channel etc.) in the corresponding cell
ABS scheme:
● Time-domain technique
● Given an ABS ratio (i.e. ratio of blank over total #subframes), a user may select a cell with
maximum ABS ratio (typically 10 to 20 %)
● Improves overall throughput of the off-loaded users by sacrificing throughput of unbiased
BS

22. Cell Range Expansion (CRE) – an ICIC mechanism
Thursday, 24 August 2017 22
High-power
Macro cell
Low-power
small cell

23. Almost Blank subframes (ABS) – an eICIC mechanism
Thursday, 24 August 2017 23

24. User Association: Main Modelling Approaches
Thursday, 24 August 2017 24
v Based on “Utility” modelling
v Example of Utility functions: spectrum efficiency , energy efficiency, QoS, outage/blocking
ratio, fairness,…
v Main Approaches:
Ø Game Theory (interaction of multiple players (users and cells) until equilibrium)
Ø Combinatorial Optimisation (utility maximisation under constraints)
o Well suited as a centralised approach hence allowing deployments at a light data
centre at the edge of coverage
o Captures all the cells in a cluster (or network) conditions hence achieving better
resource utilisation
Ø Stochastic geometry (captures topological randomness of the network geometry)
[1] D.Liu et al, “ User Association in 5G Networks: A Survey and an Outlook”, IEEE communications Surveys & Tutorials, Vol.18, No. 2, Second Quarter 2016
[2] A. Mesodiakaki, F. Adelantado, L. Alonso, and C. Verikoukis, “Energy efficient context-aware user association for outdoor small cell heterogeneous networks,” in Proc. IEEE Int.
Conf. Commun. (ICC), Jun. 2014, pp. 1614–1619.
[3] S. Corroy, L. Falconetti, and R. Mathar, “Dynamic cell association for downlink sum rate maximization in multi-cell heterogeneous networks,” in Proc. IEEE Int. Conf. Commun.
(ICC), Jun. 2012, pp. 2457–2461.
[4] H. Zhou, S. Mao, and P. Agrawal, “Approximation algorithms for cell association and scheduling in femtocell networks,” IEEE Trans. Emerging Topics Comput., vol. 3, no. 3, pp.
432–443, Sep. 2015.
[5] R. Madan, J. Borran, A. Sampath, N. Bhushan, A. Khandekar, and T. Ji, “Cell association and interference coordination in heterogeneous LTE-A cellular networks,” IEEE J. Sel.
Areas Commun., vol. 28, no. 9, pp. 1479–1489, Dec. 2010.

25. Shared Spectrum Issues for Small Cells
Thursday, 24 August 2017 25
• Extremely high cost and scarcity of dedicated licensed spectrum bands.
• Efficient use of spectrum in 5G networks will rely on sharing rather than exclusive licenses to ease congestion
in licensed bands and to increase capacity.
• Methods for mutually acceptable sharing strategies include looking up a central database with current
location to find the permitted frequencies, RF power levels etc. (SON self planning/configuration issue)
• In Co-primary Spectrum Sharing (CoPSS) , any operator is allowed to use shared spectrum. Primary license
holders agree on the joint use of (or parts of) their licensed spectrum.
• Suitable for small cells especially when base stations have a limited coverage similar to that of WiFi access
points and the frequency is dedicated to small cell use.
• LTE technology alternatives in unlicensed spectrum include : LTE WiFi aggregation (LWA), LWA using IPSEc
Tunnel (LWIP), LTE Licensed Assisted Access (LAA) and LTE in the unlicenced spectrum (LTE-U)
[1] P. Luoto et al., “Co-primary multi-operator resource sharing for small cell networks,” IEEE Trans. Wireless Commun., vol. 14, no. 6, pp.
3120–3130, Jun. 2015.
[2] Intel White paper, “Alternative LTE Solutions in Unlicensed Spectrum: Overview of LWA, LTE-LAA and Beyond, 2016

26. Some Challenges and Open Research Issues
Thursday, 24 August 2017 26
• Signalling overheads /Latency constraints/backhaul constraints
• Complexity and implementation issues
• Energy efficiency
• Multi RAT operation
• Context awareness solutions /prediction of user behaviour

27. Summary
Thursday, 24 August 2017 27
• Small cells are one of the main solutions to 5G capacity targets
• SON techniques are key to efficient deployment, optimisation and
operation of dense small cells.
• Many open questions and research challenges remain to be
investigated.

28. Thanks for your attention !