1. Celcom KV/OKV Hammer Project
7/17/2018
Ericsson KV/OKV Project Transport Network LTE/3G Backhaul & FTV Analysis
Transport Network – February 2017
© Ericsson AB 2017

2. LTE Infrastructure topology
Celcom KV/OKV Hammer Project
7/17/2018
LTE Infrastructure topology
S1 Interface is between the eNB and EPC
It’s divided into the
control plane and
user plane.
The S1 control plane terminates in the MME core network node
The S1 user plane terminates in the SGW core network node
Both are carried over IP Backhaul, So any issue in the IP Cloud would have an impact on all layers.
User Data plane uses UDP which is very efficient in Time-Sensitive Platforms, but depends on Control Plane for Reliability and insurance of delivery
© Ericsson AB 2017

3. Celcom KV/OKV Hammer Project
7/17/2018
3G/LTE Infrastructure topology
The control plane which uses SCTP, controls the delivery aspect & redundancy to insure the proper transport functionality
Therefore control plane (SCTP) has Higher priority over user plane
If Control Plane (SCTP) itself is suffering from transport/connectivity issues then whatever it controls will suffer as well, this will cause discarded and retransmitted data packets and will impact end-user performance
IuB in 3G is the equivalent to S1 in LTE.
IuB is the whole Path (links) between RBS to RNC
Commonly 3G & LTE Share the Backhauling Transmission
SCTP Are sent in CHUNKS, which a sort of Container which carry several Packets.
This container Size It must be aligned through the whole NW Nodes to insure a proper delivery of Service level
© Ericsson AB 2017

4. Celcom KV/OKV Hammer Project
7/17/2018
Study Guide
The findings of the Study conducted on the NW is done from Access/RF Observability Point of view.
No data is provided from IP/MPBN/MPLS Monitoring Parties.
Using different data collection methods:
IuB Performance Statistics for 3G.
SCTP Chunks Performance Statistics for LTE
TWAMP Probe Readings
Statistical Data collected are for the period between 1st Nov till 7th Feb
© Ericsson AB 2017

5. Celcom KV/OKV Hammer Project
7/17/2018
Nation Wide View (KV & OKV)
The Balloons Highlighted in the Maps are not just indicating the NW sites.
In fact they are sites with sever TN Issues.
Indeed they are to much, but this how the TN is effecting the NW performance.
Criteria is as follow:
Statistics of 3 Months daily Peaks Average.
LTE sites with SCTP loss Ratio ≥3%
3G Sites with IUB Limitation ≥ 10%
© Ericsson AB 2017

6. LTE Performance (KV)
In KV More than 70% of the Sites suffers High SCTP Loss ≥1%, which is Very High for Control Plane.
An Action/Tune Was done – Timer or Size – in either PS Core (MME/SGSN) or Transport NW.
This action reduced the Number of CHUNKs sent which reliefed the NW and as an effect the Data Traffic Increased Dramatically.
Yet SCTP Loss Ration Increased, That is an indicator that regalrdless of the action there are still Packets that are Lost due to Limitation and Botlenecks.

7. 3G DL Performance (KV)
In KV More than 40% of the Sites suffers High IuB Limitation ≥5% this limitation will directly impact throughput.
A continues Limitation should be considered a serious issue specially with Growing Multi-Tech Networks
The Whole Path should be considered for Upgrade Not just Singla Sites Link, this include Aggregation Points/HUB/Switches/Routers.
Coming slides will explain more the scope and impact of this limitation.

8. 3G UL Performance (KV)
Even though Few sites suffer Sever UL IUB Limitation In KV (3% Of Sites, ≥5% limitation) it never considered Normal as this is an indication of something very wrong.
Common Transport Planning Consider DL bandwidth either same or few times Higher than UL bandwidth, this is because DL Traffic is always higher in many magnitude than UL.
It should be noted that it will also impact DL Throughput as request will be delayed causing Ping and long waiting for DL responses.
But with growing Demand for many UL services traffic is increasing in UL which insure the Need for better TN Plan/Expansion/Upgrade to coup with this Trend.

9. LTE Performance (OKV)
In OKV More than 90% of LTE Sites suffers High SCTP Loss ≥1%.
With On-going Modernization and Increase of Users, the NW is being overloaded.
From Trends it seems many actions are taken to ease the impact on the TN.
But until it’s full Transport Upgrade\Optimization some cases will make other part degrade causing bad user experience (DL Throughput).

10. 3G DL Performance (OKV)
In OKV More than 30% of the Sites suffers High IuB Limitation ≥5% this limitation will directly impact throughput.
Above graphs shows when DL Limitation decreased immediately the DL Throughput Improved.
Still the average limitation value is considered very high for NW Level.
Coming slides will explain more the scope and impact of this limitation.

11. 3G UL Performance (OKV)
In OKV it’s different Story as 10% Of Sites suffers ≥5% UL IUB limitation This is an extreme case.
In the Coming slides will it shows that SABAH & SARAWAK has the Highest share of those sites, this explain the Long Ping delays and Bad response experience.
Also - as mentioned before – It will impact the User Experience and DL throughput as will.

12. Celcom KV/OKV Hammer Project
7/17/2018
Transmission Aggregation Nodes
Aggregation Nodes are smaller IP backbone points:
They could be a HUB Site
Or a dedicated Switch/Router HUB
Any issues with the Nodes throughout the Path will impact the Whole Path and all other Paths connected to that Node.
As will be shown in the following Maps sites Highlighted are suffering S1 SCTP Loss And/or IuB Limitation, and always seem to follow a defined Path with assumed Multiple aggregation points
from Radio/Access Point of View S1/IuB is the whole Path from Site-to-Core.
In the next slides we will present how suffering sites are distributed in the area each area.
Criteria of selecting sites are as follow:
Statistics of 3 Months Daily Peaks Average.
LTE sites with SCTP loss Ratio ≥3%
3G Sites with IUB Limitation ≥ 10%
A HUB with Bandwidth issues will be impacting 3G/LTE because they share the same backhauling
Usually 3G will be more indicative to Transport Bottlenecks because LTE has higher priority But it will be effecting
© Ericsson AB 2017

13. Celcom KV/OKV Hammer Project
7/17/2018
Transmission Aggregation Nodes
KV - KPRN31
In KV – KPRN31 which is the Area with most sites suffering High IUB limitation/SCTP Loss Ratio.
As shown in the Map as an assumption – Needs TN Team Confirmation - It seems there are two redundant paths,
one alongside the sea and.
The other within the land.
The Concentration of LTE sites indicates a Bandwidth issue in the HUB
© Ericsson AB 2017

14. Celcom KV/OKV Hammer Project
7/17/2018
Transmission Aggregation Nodes
KV – SHRN45
In KV – SHRN45 which is the smaller area in KV, sites suffering High IUB limitation/SCTP Loss Ratio seems to be due to one of the Main HUBs.
As shown in the Map two paths shows Limitation all heading towards the same concentration point.
© Ericsson AB 2017

15. Celcom KV/OKV Hammer Project
7/17/2018
Transmission Aggregation Nodes
KV – SHRN46
In KV – SHRN46 area which is as biggest in Terms of Traffic, sites suffering High IUB limitation/SCTP Loss Ratio seems to be due to multiple links capacity and aggregation points.
As shown in the Map two paths shows Limitation one heading towards the north and the other slightly to the south of the Concentration area.
© Ericsson AB 2017

16. FTV – OKV (3G/LTE) In 3G KV In LTE KV
50% of the Cells falls below the 15Mbps level
90% of the Cells falls below the 25Mbps level
Majority of 3G Sites Speed is on level of 1 Carrier throughput Peak which is 21Mbps.
This is on a netwrok were A 1 User practical Peak throughput Should be more than 35Mpbs on MulitCarrier (Theoretical 42Mbps)
In LTE KV
50% of the Cells falls below the 42Mbps level
90% of the Cells falls below the 70Mbps level
With LTE sharing same Backhauling with 3G But being Prioritized, 50% of LTE Sites are on 3G (MultiCarrier) Peak Throughput Ranges
On 90% Percentile they barely reach the Average LTE 10 MHz Throughput, 70Mbps

17. Celcom KV/OKV Hammer Project
7/17/2018
Transmission Aggregation Nodes
OKV – EASTERN
In OKV – EASTERN Region, sites suffering High IUB limitation/SCTP Loss Ratio are concentrated within the Main Cities and along some paths
Distant sites sometimes suffer limitation/Loss, in many cases this would be due to limitation on Site-to-Site Link Status/Capacity not just Site-To-Controller.
It’s very clear that a TN Revision Expansion and Optimization is must to solve this issue as Traffic increase will more cause degradation in user Experience and bad service quality.
© Ericsson AB 2017

18. FTV – OKV (EASTERN LTE) In EASTERN
50% of the Cells falls below the 42Mbps level
90% of the Cells falls below the 58Mbps level
the deviation is the lowest, with better Throughput than other OKV Regions.

19. Celcom KV/OKV Hammer Project
7/17/2018
Transmission Aggregation Nodes
OKV – SABAH
In KV – SABAH Region has a very unique IUB limitation Condition, The IUB Limitation in UL – which usually is A very rare case – is higher than DL in many Sites as seen in the map (Green Labels)
IUB Limitation/SCTP Loss Ratio is Concentrated in the Main Cities, specially the West Side.
SABAH is worst region Suffering UL TN Issues But of course DL will be impacted as well.
© Ericsson AB 2017

20. FTV – OKV (SABAH LTE) In SABAH
50% of the Cells falls below the 40Mbps level
90% of the Cells falls below the 60Mbps level
In this region the deviation is a little worse than Eastern,
But It’s noticed that Throughput start from Higher Ranges,
this could be an indication of Backhaul limitation as Throughput distribution start High yet not exceeding a certain limit compared to other OKV Regions

21. Celcom KV/OKV Hammer Project
7/17/2018
Transmission Aggregation Nodes
OKV – SARAWAK
In KV – SARAWAK Region like in SABAH many sites has an IUB Limitation in UL.
SARAWAK is ranked the worst region in Terms of number of Sites with TN Issues, Almost All the Sites are suffering.
© Ericsson AB 2017

22. FTV – OKV (SARAWAK LTE)
In SARAWAK
50% of the Cells falls below the 28Mbps level
90% of the Cells falls below the 47Mbps level
In this region the deviation is the worst, and throughput is also the lowest
Sites are mostly concentrated around the 50% (28Mbps), another clear sign of Backhaul bottleneck issues

23. TWAMP - Two-Way Active Measurement Protocol, Responder
This feature enables the Site (BB) to act as one or more logical TWAMP Light Responders with state full test session management. It enables the computation of the one-way packet loss, reordering, and duplicate metrics while ensuring interoperability with different active monitoring probe vendors including Ericsson.
The Ericsson IP Probe system can measure the following metrics on UDP Layer:
Round-trip packet delay
One-way packet delay
One-way packet delay variation
One-way packet loss
One-way packet reordering
One-way packet duplication
An Important part is that usually the Probe is installed in Main aggregation points/router/switches.
A whole single path from site to the Core could be measured to detect Transport issues.
It will be difficult to detect exact location of smaller HUBs bottlenecks in the Midway.
unless another Probe is installed but that would be inefficient and would overflow the Transport with Test Packets.
That’s why Other Statistical data is used to Analysis the Transport Network Performance.

24. TWAMP Monitoring
For sites needs to be checked the Responder is activated then the Test packets start flowing measuring the performance of the Path.
For a specified period a summery table and trending graph for each activated site/Path is generated.
The measured metrics are (P=Probe, R=RBS):
P2R (DL) & R2P (UL) packet loss ratio (Critical)
P2R (DL) & R2P (UL) packet delay (Important)
P2R (DL) & R2P (UL) packet delay variation (Indicative)
Sample Graphs Next Slides.

25. TWAMP – Sample Site (KV)
(BB1_B058D_B00376OD_PLBHBARAT2)
P2R (DL) packet loss ratio is High, ≥0.0% off-Peak, ≥5% Peak
R2P (UL) packet loss ratio is almost null
On the other Hand the PDV (packet delay variation) in R2P (UL) is comparatively very high. This is indicative of routing issues or potential bottlenecks

26. TWAMP – Sample Site (OKV)
(BB1_Q136A_Q00387OD_SARATOK)
P2R (DL) packet loss ratio is High, ≥10% off-Peak, ≥30% Peak
R2P (UL) packet loss ratio is almost null
PDV (packet delay variation) in P2R (DL) is varying but not much. But it’s matching the Loss Ratio, This indicate a link Capacity issue