Power Consumption of IPv6 vs IPv4 in Smartphone
CEA, LIST
CEA Saclay
Gif-sur-Yvette
Ile-de-France
91190
France
+33169089223
Alexandre.Petrescu@cea.fr
CEA, LIST
CEA Saclay
Gif-sur-Yvette
Ile-de-France
91190
France
+33169082939
Siwar.Benhadjsaid@cea.fr
Greenspector
Loire-Atlantique
France
ophilippot@greenspector.com
Greenspector
Loire-Atlantique
France
tvincent@greenspector.com
Operations and Management Area
IPv6 Operations Working Group
IPv6, IPv4, power consumption, battery
This draft documents preliminary results of measuring the
power consumption of using IPv6 vs using IPv4 on typical
applications on a smartphone. The smartphone is connected on
a 4G cellular network with either an IPv6 connection, or with
an IPv4 connection, but not both simultaneously.
The preliminary results expose a roughly 5% increase in power
consumption on IPv6. More experiments are planned as future
work that may confirm or infirm these figures.
This document describes preliminary measurement results of
power consumption of IPv6 and IPv4 on a cellular smartphone.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119.
APN: Access Point Network
We conducted several power consumption in differents use cases. The
measurement are done with Greenspector probe. We used the
smartphone Huawei mate 8 with an android version 7. In these tests,
we used orange SIM card. The smartphone is able to connect to IPv4,
or to IPv6 Access Point Networks (APNs) simultaneously, or
distinctively. In the smartphone, only the 4G connection is active
(not 3G).
The test methodology consist in an automatized test and a
probe provided by Greenspector. First, we specify in the
automatized test which application we want to monitor and for
which duration. Then, we trasfer the test and the probe to
the smartphone, launch it and wait till the end of the test.
The probe generates a JSON file where measures are stored. We
take out the JSON files from the smartphone and send them to
Greenspector server in order to be anlyzed. Finally,
different metrics (e.g. battery discharge, process CPU,
network packets, etc.) are shown in a Greenspector webpage
(access available upon request). Some of the measurements are
graphically illustrated in a Digital Object; this object is
identified by DOI 10.13140/RG.2.2.13454.77122. This
Identifier can be resolved by a publically available resolver
such as dx.doi.org.
All the measurements presented in this draft are preliminary
and may change depending on the details of the measurement
conditions.
In order to compare between IPv4 and IPv6 connections in smartphone,
we considered four use cases: video streaming from YouTube, live
video streaming (no cache), web browsing and file download. First,
we run the test, for each of the four use cases, when the smartphone
is using 4G connection and IPv4 APN. Then, we run the same tests
when the smartphone is using 4G connection and IPv6 APN (distinct
connection set up; in each one of these connections only IPv4 is
active, or only IPv6 active).
In this use case, we launch a video of 4m44s using Youtube
application in the smartphone. We associate the Greenspector probe
to Youtube application in order to measure the related power
consumption. The measurement results (a significant selection) are
shown in Table . As we can
see, in same environment and by moving from the use of IPv4 to IPv6,
the battery discharge is increased by around 5%.
IP version
Battery discharge
IPv4
57.61 mAh
IPv6
60.90 mAh
In this use case, we launch a live video during 5 minutes
using Youtube application. For this use case, we considered
the live streaming provided by a webcam in Amesterdam. This
webcam can be reached either using IPv4 or IPv6 connection
(https://www.youtube.com/watch?v=Y5N-OM9RT8o). It was setup
in order to demonstrate that it is possible to use IPv6
exclusively in the production and delivery of a live video
stream (https://www.terena.org/webcam/). We associate the
Greenspector probe to Youtube application in order to
measure the related power consumption. The measurement
results are shown in . We note that, in IPv6
network, the streaming of a live video in smartphone
consumes approximately 3% more than IPv4 network.
IP version
Battery discharge
IPv4
66.11 mAh
IPv6
67.89 mAh
In this use case, we launch Chrome application with the URL
that can be browsed using both of IPv4 and IPv6 connections
(e.g. http://www.kame.net). We associate the Greenspector
probe to Chrome application in order to measure the related
power consumption. The measurement results are shown in
. We note that
browsing using IPv6 networks discharges the battery
approximately 12% more than browsing using IPv4 networks.
IP version
Battery discharge
IPv4
3.31 mAh
IPv6
3.73 mAh
In this use case,we use the "andFTP" application to
download a file of 7MBytes. The "andFTP" is an android
application dedicated for file transfer. We associate the
Greenspector probe to the andFTP application in order to
measure the related power consumption. The measurement
results are shown in . We
note that downloading a file using IPv6 network discharge
the battery approximately 36% more than using IPv4
network.
IP version
Platform discharge
IPv4
69.21 mAh
IPv6
94.15 mAh
In the next revisions of this draft, the following topics will
have to be addressed:
Measure consumption of HTTP2 protocol, compare IPv4 and
IPv6. An earlier measurement of HTTP2 is available at
https://greenspector.com/articles/article-http2-latency-and-energy.html
Use other smartphone brand and OS version.
Analyse the wireshark traces to understand whether or not
the right parameters are used for IPv6.
Use an alias in order to make sure be on same POP, and
then try with different POPs.
Perform measurements and compare power consumption of WiFi
IPv4 vs WiFi IPv6.
There is a need to explore how much more power consumption (if
any) is generated by https vs http, with IPv4 and IPv6.
This work is supported by the European Celtic-Plus project
CONVINcE and was partially funded by Finland, France, Sweden
and Turkey.
This work was performed with SIM cards on Orange IPv6 4G
cellular network deployed in France; Patrick Ainard Simonet.