Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Public
5G coverage and performance assessment in UK top cities and Combined Authorities
Customer: West Midlands 5G 11th December 2020
Public
Picture
Contents.
5G Coverage
5G Throughput
5G Latency/Data Availability
Appendix: Methodology
Public
Data Collection.Data analysis of samples collected over a 4-month period.
311-Dec-20
Data collection period: July 20th (wk 30) – November 29th (wk 48)
KPI Definition
Coverage – 5G%% of tiles covered with 5G. The tile size is ≈ 100m x 100m. For a tile to be considered as covered at least 1 5G (NR)
sample is required.
Throughput Avg & P95 of throughputs from 5G capable users. A user is considered 5G capable when he has reported more
than 5 5G samples in a week
Data Availability % of successful connection tests when device is on 5G
Latency Avg latency when device is on 5G
Data collection methodology details per KPI are shown in slides 21-32.
Devices having contributed 5G data in the time period and polygons of interest
Device Name % of total samplesSG Note 10 24.5%SG S20 5G 22.7%
SG S20+ 5G 17.9%SG S20 Ultra (5G) 14.0%SG Note20 Ultra 6.1%
SG S10 5G 5.1%SG A90 5G 2.8%
SG Fold (5G) 1.9%Sony Xperia 1 II 0.8%
SG Note20 0.8%OPPO Find X2 Pro 0.6%
SG A51 5G 0.5%Huawei Mate 20 X (5G) 0.5%
SG S20 FE 5G 0.4%Motorola moto g 5G plus 0.2%
Pixel 5 0.2%Sony Xperia 5 II 0.2%
Oppo Find X2 Lite 0.2%SG Z Fold2 5G 0.1%
Others 0.5%
5G coverage and performance assessment
Public
5G Coverage.
Public
Combined Authorities: 5G%.West Midlands ranks 1st in 5G geographical coverage.
511-Dec-205G coverage and performance assessment
Public
Cities: 5G%.Birmingham ranks 4th in 5G geographical coverage.
611-Dec-205G coverage and performance assessment
Public
City maps: 5G%.Tiles with 5G coverage are shown in blue.
711-Dec-20
Manchester Liverpool
5G coverage and performance assessment
Public
City maps: 5G%.Tiles with 5G coverage are shown in blue.
811-Dec-20
Glasgow Birmingham
5G coverage and performance assessment
Public
City maps: 5G%.Tiles with 5G coverage are shown in blue.
911-Dec-20
Bristol London
5G coverage and performance assessment
Public
Combined Authority maps: 5G%.Tiles with 5G coverage are shown in blue.
1011-Dec-20
West of England Cambridgeshire and Peterborough
5G coverage and performance assessment
Public
Combined Authority maps: 5G%.Tiles with 5G coverage are shown in blue.
1111-Dec-20
West Midlands Greater Manchester
5G coverage and performance assessment
Public
Combined Authority maps: 5G%.Tiles with 5G coverage are shown in blue.
1211-Dec-20
Liverpool City Region North East
5G coverage and performance assessment
Public
Combined Authority maps: 5G%.Tiles with 5G coverage are shown in blue.
1311-Dec-20
North of Tyne Sheffield City Region
5G coverage and performance assessment
Public
Combined Authority maps: 5G%.Tiles with 5G coverage are shown in blue.
1411-Dec-20
Tees Valley West Yorkshire
5G coverage and performance assessment
Public
5G Throughput.
Public
Cities: DL/UL Throughput.Birmingham ranks 2nd in 5G DL/UL throughput behind Liverpool.
1611-Dec-205G coverage and performance assessment
Public
Combined Authorities: 5G Throughput.Users in West Midlands experience 60% DL thr. improvement in their daily activities when connected on 5G.
1711-Dec-205G coverage and performance assessment
Public
5G Latency/Data Availability.
Public
Cities: 5G Data Availability and Latency.
1911-Dec-205G coverage and performance assessment
Public
Combined Authorities: 5G Data Availability and Latency.
2011-Dec-205G coverage and performance assessment
Public
Methodology.
Public
Methodology: Combined Authorities polygons.Polygon set for the Combined Authorities.
225G coverage and performance assessment 11-Dec-20
Combined Authority polygons were extracted from:
https://data.gov.uk/dataset/d2094bbf-8c63-4218-95df-92e870b1eecc/combined-authorities-march-2017-full-clipped-boundaries-in-england
The latest available government polygon set is from 2017 and thus includes outdated boundaries for the North East combined authority. The newly established combined authorities include the following councils:
• North of Tyne: Newcastle, North Tyneside and Northumberland
• North East: Durham, Gateshead, South Tyneside and Sunderland
Council polygons were extracted from:
https://data.gov.uk/dataset/11302ddc-65bc-4a8f-96a9-af5c456e442c/counties-and-unitary-authorities-december-2016-full-clipped-boundaries-in-england-and-wales
Public
Methodology: Cities polyogns.Polygon set for the top 6 cities.
2311-Dec-20
City polygons for England:
Birmingham, London, Manchester, Liverpool and Bristol
https://data.gov.uk/dataset/7879ab82-2863-401e-8a29-a56e264d2182/major-towns-and-cities-december-2015-boundaries
Polygon for Glasgow:
https://data.gov.uk/dataset/1a9d2abf-9050-42ac-b508-02cf2f4532b2/localities-urban-areas
5G coverage and performance assessment
Public
Coverage
Public
Coverage.GPS location information
Network Information use case
⚫ Network Information result use case is triggered after every GPS update from the operating system. For example, during a Google Maps navigation.
⚫ Our apps only trigger GPS during an active voice call to get exact location in case the call drops.
⚫ It is the use case with highest density of samples due to the fact that each sample is triggered every time there is a GPS update on the phone
⚫ Samples collected from both the Network Information and Connection Test use cases are included in the coverage analysis.
25
Network Information samples are produced every second, when GPS is enabled and updated.
The app with the umlaut SDK does not have to run actively. Installation is sufficient.
t
Trigger events:
1 GPS: Location update
5G coverage and performance assessment 11-Dec-20
Public
Throughput
Public
Network Traffic/Throughput Result (NTR).Passive Data Collection
⚫ Calculate and store the “fastest representative value” throughput rate within the 15min interval⚫ per traffic direction (UL/DL) and⚫ per technology (2G/3G/4G/5G/WiFi)
⚫ The used aggregation of the per-second throughput rate depends on the statistics (number of seconds) accumulated within the 15 minutes for each technology
⚫ The “fastest representative value“ is represented by a measurement point from the top data rates to avoid limited throughputs and outliers but always trying to keep the statistical re-semblance, calculated as follows:
Description
27
Filtering
Number of samples Percentile
< 10 No Throughput considered
[10 – 20] P80
[21 – 50] P95
> 51 P99
⚫ Passive collection of user experienced throughput is monitoring user activity
⚫ A big proportion of throughput samples are not reflecting the network capabilities as they are derived from less demanding data sessions (i.e. WhattsApp messaging)
⚫ As such, an analysis taking into account all throughput samples will result in artificially low values and mask any network performance differences
⚫ Filtering to filter out data sessions without sufficient data traffic:
Radio Access Technology Data Traffic Filter
2G Traffic bytes >= 500,000
3G Traffic bytes >= 2,000,000
4G & 5G Traffic bytes >= 10,000,000
5G coverage and performance assessment 11-Dec-20
Public
Data Availability
Public
Connection Test.Internet availability and accesibility
⚫ Download a small icon (596 bytes) via
TCP/SSL.
⚫ Test is executed every 15 min in the
background.
⚫ The test result is either successful,
representing mobile state internet state
(cellular or WIFI), or failed.
⚫ Connection test 4 stages:
⚫ DNS lookup
⚫ TCP Connect
⚫ HTTP received
⚫ Overall success
Connection Test description: Connection Test Sequence:Client
DurationDNS
DurationTCP connect
DurationSSL handshake
HTTPs GET
DurationHTTPs receive
DNS Server
Target:Multi-CDN
Server
Prox
yPr
oxy
DNS Lookup
DNS Response
SYNSYN /ACK
ACK
SSL Client Hello
SSL Initialization
HTTPs GET (HTTP 1.1)
HTTPs Receive
11-Dec-20 295G coverage and performance assessment
Public
Connection Test.Internet availability and accesibility
Restrictions:
⚫ Battery level must be higher than 15%.
⚫ Device / host app must not be in battery saving or
doze modes.
Metrics The test returns several metrics, e.g. :
⚫ DurationDNS: DNS lookup time in milliseconds
⚫ DurationTcpConnect: TCP connect time in milliseconds
⚫ DurationHttpReceive: Duration of file content transfer in
milliseconds
⚫ Success Per test stage; TRUE if test stage was successful
305G coverage and performance assessment 11-Dec-20
Public
Latency
Public
Latency Test Result.
11-Dec-20 32
Latency measures
⚫ A latency test is executed after every successful connection test (CT). This is perform by doing 10 pings to the Amazon or Akamai CDN addresses.
⚫ It is collecting the round trip time (RRT) for each of the pings and the results are the min, max, avg, med and jitter values of the pings
⚫ Latency Results with minValue < = 0 or maxvalue > 60000 are filtered out and also those where the number of successful pings does not match the number of executed pings. As such, only samples where each ping was successful and gives a reasonable latency value are used for subsequent analysis
⚫ Latency tests that started more than 3 seconds after Connection Test are also discarded
Description:
Considerations:
Sequence Schematic :
Delta
Connection Test (CT)
DNS
Latency Test (LTR)
LTR Timestamp onStart
Run 10 pings with 200 ms pause, 30 s timeout
(ping -i 0.2 -c 10 –W 30 …)
… … …i
RTT 1 RTT10
Close/ Write Log
LTR Timestamp onEnd
…
5G coverage and performance assessment
Public
Key contacts for further information.
Hakan Ekmen
+49 151 [email protected]
Velissarios Lataris
+44 7758 [email protected]
Efstathios Tsardinakis
+44 7398 [email protected]
11-Dec-205G coverage and performance assessment 33
Public
www.umlaut.com
3411-Dec-205G coverage and performance assessment
Public 35
Disclaimer
This document and all information contained herein is the sole property of umlaut.No intellectual property rights are granted by the delivery of this document or the disclosure of its content. This document shall not be reproduced or disclosed to a third party without the express written consent of umlaut. This document and its content shall not be used for any purpose other than that for which it is supplied.
11-Dec-205G coverage and performance assessment