ran dimensioning: lessons learned by telstra
TRANSCRIPT
June
1CLIENT CONFIDENTIAL
Copyright © Maravedis-Rethink 2015
RAN dimensioning: Lessons learned by Telstra
April 30, 2015
Copyright © Maravedis -Rethink 2015
• Caroline Gabriel, Research Director, Maravedis-Rethink
• Mr. Max Downey, SME Capacity Management, Telstra
• Dr. Bruce Northcote, CEO, Telari
• Q&A
AgendaApril 2015
June
3CLIENT CONFIDENTIAL
Copyright © Maravedis-Rethink 2015
The capacity challenge moves on to customer experience
Caroline Gabriel, Research Director, Maravedis-Rethink
CLIENT CONFIDENTIALCopyright © Maravedis-Rethink 2013
www.maravedis-bwa.com
4
Dense networks to meet rising demand, but capacity alone does not create profit
The RAN capacity explosion
April 2015
Copyright © Maravedis Inc 2015.
www.maravedis-bwa.com
• Capacity crunch and revenue crunch –a new approach to the RAN is essential• New ways to increasing real capacity –hardware capex heavy and a law ofdiminishing returns• New focus on optimization andresource management• Soft, dynamic capacity key to survivalof the MNO model
Two key differences from the past:
• More elements to optimize• Different objectives driven by CEM
Growth in mobile broadband capacity investment relative to incremental revenues for a typical developed market MNO(relative to capex baseline of ‘1’ in 2013)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2013 2014 2015 2016 2017 2018 2019
Gro
wth
re
lati
ve t
o c
apac
ity
= 1
Capacity investments Data revenues
CLIENT CONFIDENTIALCopyright © Maravedis-Rethink 2013
www.maravedis-bwa.com
5
Need to dimension and optimize many types of cell equipment in dense networks
The new RAN – danger of chaos?
April 2015
Copyright © Maravedis Inc 2015.
www.maravedis-bwa.com
% of carriers adopting each of the key architectures as their primary approach , MaRe January 2015
0
10
20
30
40
50
60
70
80
2014 2015 2016 2017 2018 2019
% o
f ca
rrie
rs a
do
pti
ng
this
ap
pro
ach
as
pri
mar
y
Macro-first Carrier WiFi Small cell HetNet layer RAN virtualization
OR
CLIENT CONFIDENTIALCopyright © Maravedis-Rethink 2013
www.maravedis-bwa.com
6
Capacity not an end in itself, network drivers shifting to real world experience
The new KPIs – customer-centric RANs
April 2015
Copyright © Maravedis Inc 2015.
www.maravedis-bwa.com
• Fewer operators measure RAN successjust with the three Cs – cost capacitycoverage• Hardware cost considerations shift toTCO and ROI, including software• QoE key competitive differentiator,features in two of top 5 network KPIs• Poor network QoE – second biggestreason for churn after cost• QoE may also support higher ARPU, newtypes of service eg enterprise, safety• Need to distinguish high value customersor apps
KPIs named as most important for RAN (top 5 identified in CTO survey, then respondents selected most important of those) Source MaRe survey March 2015
27
2518
16
14
Lower TCO
QoE to reduce churn
Analytics
QoE to support newrevenues
Capex reduction
CLIENT CONFIDENTIALCopyright © Maravedis-Rethink 2013
www.maravedis-bwa.com
7
Large amounts of capacity locked up by static architectures
Key ways to achieve best experience
April 2015
Copyright © Maravedis Inc 2015.
www.maravedis-bwa.com
Significant ways to make more from current networks, and plan new ones
• Unlock soft capacity and underused resources• Map capacity to demand by time, location, app• Dimensioning as part of broad optimization picture• Link RAN data to OSS/BSS• Dynamic resource allocation
Results
• Cost savings• Deferred hardware capex• Higher customer satisfaction• Lower churn• Support for new business cases
CLIENT CONFIDENTIALCopyright © Maravedis-Rethink 2013
www.maravedis-bwa.com
8
Shift in emphasis from sites to customer experience in real terms
New priorities in capacity management
April 2015
Copyright © Maravedis Inc 2015.
www.maravedis-bwa.com
From this To this
TELSTRA T
EM
PLATE 4
X3 B
LU
E B
ETA |
TELPPTV4
RAN DIMENSIONING
When investing CAPEX in capacity related RAN infrastructure, we must try to
balance a number of competing goals:
• Maximise the customer experience
• For the largest number of customers
• With the most effective use of invested capital
• Through the prioritised expansion of network hardware driven by business needs.
Pr
es
ent
ati
on
Titl
e |
Pr
es
ent
er
Page 10
SO, WHAT’S THE POINT?
SO, WHAT’S THE POINT?
Budget
($$$)
Scenario Modelling
Devil Engineers
Business Requirements
(… or, what are we buying?)
$
Reporting/Forecasting
Scenario List
We want to be able to know how
much a certain level of network
performance/customer
satisfaction will cost.
Accurate reporting closes the feedback loop
allowing us to both:
• Describe how specified ‘Business
Requirements’ will affect network
performance
• Validate forecasts by monitoring
actual network performance.
Translate business
requirements into capacity
actions.
• Maximise the customer
experience
• For the largest number of
customers
• With the most effective use of
invested capital
• Through the selective expansion
of network hardware driven by
business needs.
DIMENSIONING CONTROLS
Un-optimised Optimised
$
Capacity
Modelling
The setting of Network Controls needs to be translated into a set of required actions
(what should be built and where, etc.).
The impact of these network actions needs to be modelled, so that future network
performance can be forecast.
There are two main types of modelling/analysis that drive this stage of the capacity
process:
• Intrinsic Analysis (automatic, based on well defined rules)
• Extrinsic Analysis (qualitative, based on human interpretation of business
goals and local knowledge)
It’s important to have well defined and documented business rules and processes
that define how CAPEX should be prioritised. The rules are interpreted and
implemented by human engineers.
Page 13
WELL DEFINED PROCESSES
Extrinsic Prioritisation
Intrinsic Prioritisation Capacity Engineers
Revised
IT’S GETTING HARDER
Complexity is increasing:
- AMPS – just voice, Erlang-B
-GSM – Voice and GPRS/EDGE
-WCDMA – CS12, CS64, PS-R99, PS-HSDPA, PS-EUL
-LTE – Data, VoLTE, ViLTE, Over the top (QCI ???)
-5G – Internet of things, M2M -Terrifying
14PRESENTATION TITLE | PRESENTER NAME | DATE |
THE AMOUNT OF TRAFFIC WE'RE CARRYING IS INCREASING
15PRESENTATION TITLE | PRESENTER NAME | DATE |
[Cisco]
THE NUMBER OF NETWORK ELEMENTS ARE INCREASING
• HetNets
• PicoCells
• 5G millimetre Wave (+100GHz)
• Excel based dimensioning tools are not scalable.
16PRESENTATION TITLE | PRESENTER NAME | DATE |
INCREASED SERVICE AND PRODUCT COMPLEXITY
• VoLTE
• ViLTE
• eMBMS
• Differentiated QoS
• Etc.
17PRESENTATION TITLE | PRESENTER NAME | DATE |
IMPORTANT CONSIDERATIONS
Finally, any good capacity model should consider the following:
• Just in time capacity
• User Experience KPIs
• Soft Capacity
18PRESENTATION TITLE | PRESENTER NAME | DATE |
Thank you
Questions?
19PRESENTATION TITLE | PRESENTER NAME | DATE |
[ ]
20Ascom Webinar, 2015 © Ascom
DIMENSIONING CAPACITY TO USER EXPERIENCE
Presenter: Dr. Bruce Northcote
DIMENSIONING CAPACITY TO USER EXPERIENCE
Ascom Webinar, 2015 © Ascom 21
What is capacity?
Current measured load
DIMENSIONING CAPACITY TO USER EXPERIENCE
Offered Load
Ca
rrie
d L
oa
d
System capacity
y=x
Zero loss
capacity
Dimensioned capacity:
Target KPI is x% loss
Sa
fe o
pe
ratin
g r
ang
e:
Ad
mis
sio
n R
eg
ion
(AR
)
Tracking HR over time
allows me to plan for a
“just in time” capacity
expansion
100% user
satisfaction
Lowest user satisfactionI have this much
spare capacity: my
headroom (HR)
Ascom Webinar, 2015 © Ascom 22
Above this load
I will not meet
my target KPIs
1. Coverage: not a capacity management KPI
2. Resource utilisation (e.g. backhaul bandwidth, downlink power, interference
budget)
3. Service availability
4. Service retention (impacted by QoS policy, congestion control)
5. Throughput: timescales matter
a. Peak user throughput (single user speed test)
b. User throughput (speed test distribution)
c. Interactive throughput (throughput measured only when the user is interacting)
KPIS: SYSTEM-BASED VERSUS USER-EXPERIENCE-BASED
User Experience KPIs
Ascom Webinar, 2015 © Ascom 23
I can dimension my network to meet any
given set of these user experience KPIs.
The KPIs then provide the reference that
defines my network capacity.
System resources are shared dynamically between multiple simultaneous
instances of a number of services.
May apply different KPIs to each service (stringent voice, lax data).
24
CAPACITY IN MULTI-SERVICE CELLULAR NETWORKS
Ascom Webinar, 2015 © Ascom
Signalling Load
Data
Lo
ad
Admission Region: all operating
points within the region should
have achieved the KPIs
Engineered capacity
Engineered capacity
Exact shape of the admission
region is unique to each cell & can
only be determined by modelling
Soft capacity: different radio conditions
provide for more or less capacity
MODELLING SOFT CAPACITY
Ascom Webinar, 2015 © Ascom 25
ARs defined by user experience KPIs on service availability, retain-ability and
interactive uplink & downlink throughput
1. Spectrum (WCDMA codes; LTE PRB)
2. Power
3. Scheduler algorithm
4. QoS Settings & Call Admission Control (CAC)
a. Allocation, Retention & Prioritisation policy (ARP)
b. Resource reservation
5. Radio environment (weather, user location, user demand & interference)
6. User limits & licencing
7. Baseband (computational processing limits)
8. Backhaul bandwidth
FACTORS AFFECTING RAN CAPACITY
Ascom Webinar, 2015 © Ascom 26
Delivering Optimized Capacity Management
Understanding current network load using network OSS data, including soft capacity and resource allocation
Visualizing available capacity or admission region with “Headroom” for different per service growth forecasts
Predicting and alerting when cells will reach capacity
Identifying capacity constraining resources
Providing configuration recommendations to maximize capacity from the existing network
Enabling Scenario analysis: the impact of different forecasts or network expansion options
Enabling Business Decisions
Just in time capital planning
Prioritization: network or regional level “top N” lists
Prioritization: revenue contribution per cell
Consistent systemized planning across business
27
INTRODUCING TEMS™ CAPACITY MANAGER
Ascom Webinar, 2015 © Ascom
Example screenshots from Ascom’s
TEMS™ Capacity Manager
[ ]
28Ascom Webinar, 2015 © Ascom
THANK YOU FOR ATTENDING!
For more information on TEMS™ Capacity Manager, visit us at www.ascom.com/networktesting