Ali Safari Khatouni Dalhousie University AM Mandalari University Carlos III of Madrid A Custura University of Aberdeen A Lutu Telefonica Research Ö Alay Simula Metropolitan M Bagnulo University Carlos III of Madrid ID: 814885
Download The PPT/PDF document "Implications of Roaming in Europe" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Implications of Roaming in Europe
Ali Safari Khatouni (Dalhousie University), A.M. Mandalari (University Carlos III of Madrid), A. Custura (University of Aberdeen), A. Lutu (Telefonica Research), Ö. Alay (Simula Metropolitan), M. Bagnulo (University Carlos III of Madrid), V. Bajpai (Technische Universität München), A. Brunstrom (Karlstad Universitet), J. Ott (Technische Universität München), M. Mellia (Politecnico di Torino), G. Fairhurst (University of Aberdeen)
1
Slide2Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion
Slide3Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion3
Slide44
15 June 2017
https://ec.europa.eu/digital-single-market/en/roaming
Slide5Why study International Roaming?
Understand the roaming ecosystem in Europe after the “Roam like Home” initiative.
Which technical solutions are actually being deployed and used today? What are the implications of roaming on the service experienced by the roaming user? 5
Slide6Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion6
Slide77
SGW
MMEHSS
AAA
PGW
SGW
PGW
SGW
MME
HSS
AAA
PGW
SGW
PGW
eNodeB
eNodeB
Home
network
Visited
network
IPX
network
Further (mobile)
network operators
eNodeB
eNodeB
Internet
Evolved Universal
Terrestrial Radio
Access Network
Evolved Packet
Core (EPC)
Local
Breakout
Home-routed
Roaming
R
PGW
R
PGW
IPX Hub
Breakout
Mobile Node
(roaming user)
Mobile
Node
(home user)
SGW : Serving Gateway
PGW Packet Data Network Gateway
IPX: IP Packet
eXchange
Home-Routed roaming (HR) Local Breakout (LBO) IPX Hub Breakout (IHBO)
Slide8Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion8
Slide9MONROE-Roaming
nodes
9MONROE-Roaming Platform
Design Aspect
Component
Node Platform
APU2C4
Node Configuration
2xMC7455
Node Hardware
1xAPU
+
2xMC7455
Operating System
Debian
9 Stretch
Modem Type
Sierra MC7455 CAT6
miniPCIe
modem
Slide10MONROE-Roaming
nodes
MONROE-Roaming backendOne measurement server per countryMONROE-Roaming scheduler10MONROE-Roaming Platform
Slide1111
Dataset
3 months
of collected traces in
2017
and
2018
12 nodes distributed in
6 countries
16 operators (12 operators in Roaming)
More than
20000 experiments
Mobile Network Operators
NO Telia NO Telenor NO
SE Telia SE Telenor SE 3 SE
UK Vodafone UK EE
DE Vodafone DE T-Mobile O2
ES Vodafone ES Movistar Orange
IT Vodafone IT TIM 3 IT
MONROE-Roaming Platform
Slide1212
Measure one
at a time (all nodes have the SAME SIM at the same time).
Measure the visited network natively, where possible.
MONROE-Roaming Platform
[1] MNO : Mobile Network Operator
Slide13Slide14Experimental Setup
14
Slide15Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion15
Slide16Roaming Setup and Performance: Measurements
Radio metadata for tracking the roaming partner
Traceroute to discover roaming setup dig: DNS against third party service providers (ad providers)Curl: performance against 10 popular webservers16
Slide17Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion17
Slide1818
Roaming Setup and Performance:
Delay implications
99% of users experienced the RTT less than 120
ms
ECDF: Empirical Cumulative Distribution Function
Slide1919
Roaming Setup and Performance:
Delay implications
Longer the geographical distance implies the longer RTT
ECDF: Empirical Cumulative Distribution Function
Slide2020
Roaming Setup and Performance:
Delay implications
GPRS tunnel is slower than native Internet
ECDF: Empirical Cumulative Distribution Function
Slide2121
Roaming Setup and Performance:
Delay implications
Delay penalty to go back to home country
ECDF: Empirical Cumulative Distribution Function
Slide22Roaming Setup and Performance: DNS implications
For the home user the query time is significantly lower in average than for the other five roaming users
22
Home user
Roaming user
Slide23Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion23
Slide24VoIP and Content Discrimination: Measurements
Traffic differentiation measurements using three applications (FaceTime, Facebook Messenger, WhatsApp) to determine potential traffic differentiation in roamingOoniprobe web connectivity test24We do not observe any traffic differentiation on any of the 16 MNOs we measure.
We found no evidence of additional content discriminationGeo-restriction rules are the same “as home”
https://
ooni.torproject.org
Slide25Outline
Motivations
BackgroundExperimental setup: MONROE-RoamingMeasurements:Roaming setup and performanceVoIPContent discriminationRoaming resultsExperience and conclusion25
Slide26Roaming Experience
Synchronizing the measurements was the most challenging part
Synchronization over email (and sometimes phone, text, skype, smoke signals… )Re-purposing MONROE software was straightforward (thanks MONROE Engineering team!)Marvin MONROE – a scheduling daemon https://github.com/MONROE-PROJECT/SchedulerTaking care of MONROE nodes was challenging at timesNeeded intervention at the deployment site, sometimes had to re-configure the nodes at every SIM change… 26
Slide27Conclusion and Future Work
Home-Routed Roaming is the norm for the MNOs we measured
and this is going to stay there!Delay penalties on the roaming userNo traffic differentiation or content discriminationFuture work: exploration of potential performance penalties on actual end-user Quality of Experience (QoE)27
Slide28Ali.safari@dal.ca
https://www.
MONROE-project.eu/Happy Roaming to everybody!!!28
Slide29Backup Slides
Slide30Dataset
The code and the dataset collected is open to the community:
https://www.it.uc3m.es/amandala/roaming.html30
Slide31User A
Measurement setup
Video and Audio calls
.
pcap
User B
Prepare collected traces to:
Check the call establishment using the app on real mobile phone
Replay all UDP datagrams
From user A to user B (Upload link)
Slide32Measurement setup
TCPReplay
opx
Visited Country
Server
Home Country
Server
Slide33VoIP: Results
We do not observe any traffic differentiation on any of the 16 MNOs we measure.
33
Slide34Content discrimination: Results
Interference profile match Home vs. Roaming
We found no evidence of additional content discriminationGeo-restriction rules are the same “as home”34