Energy Efficiency Study of Audio-video Content Consumption

on Selected Android Mobile Terminals

Przemyslaw Falkowski-Gilski

and Maciej Pankowski

Faculty of Electronics, Telecommunications and Informatics, Gdansk University of Technology,

Narutowicza 11/12, Gdansk, Poland

Keywords: Android OS, Battery Consumption, Energy Efficiency, Mobile Devices.

Abstract: Mobile devices are widely used by billions of users worldwide. Thanks to their main advantage, which is

portability, they should be fully operational as long as possible, without the need to recharge or connect them

to external power sources. This paper describes a study, carried out on four different mobile devices,

with different hardware and software parameters, running the Android operating system. The research

campaign involved several scenarios, including consumption of audio-visual content by different means of

wireless communication (cellular and Wi-Fi), designed to best reflect the common daily use of a modern

smartphone. Those scenarios were based on a user experience survey conducted at the beginning of the study.

Obtained results illustrate user preferences as well as resource consumption of multimedia on different devices

with varying distribution of the Android OS.

1 INTRODUCTION

Nowadays, almost everyone has at least one mobile

device. The smartphone is basically a modern multi

tool. It is utilized for a wide number of applications,

e.g. communication and exchange of information

between individuals, entertainment purposes, such as

taking photos, recording videos, as well as listening

to music and consuming multimedia content

(Falkowski-Gilski and Uhl, 2020).

Mobile devices accompany us every day: at work,

at school, at home, and on the move. Everyone can

take advantage of the huge possibilities that this

pocket computer can offer (Falkowski-Gilski, 2020).

Manufacturers compete with one another in order to

create the next hit, with cutting edge integrated

technology. Yet, differences between individual

models may be either very large or negligible. Often

two separate devices have almost identical technical

specifications. Nevertheless, they differ not only in

brand, appearance, but also price. Of course, different

users have different preferences. They may vary in a

slightly different taste, needs and the way they use

their smartphones. Consequently, functionalities can

determine the choice of a user device.

https://orcid.org/0000-0001-8920-6969

The aim of this work was to test a number of

mobile devices, in terms of their energy efficiency.

We investigate the usage of resources, such as:

CPU (Central Processing Unit) and battery lifetime.

Tests, carried out during this experiment, will reflect

the typical everyday usage of a smartphone, based on

a user expectations survey.

2 USER SURVEY

The survey was carried out online using an interactive

spreadsheet application. The questionnaire consisted

of 3 questions, including both closed and open-ended

ones. Provided answers could be chosen from a

predefined list, as well as typed in by users

themselves. They were organized as follows:

1) How many mobile devices do you use every

day? (type in integer).

2) What do you usually take into account when

choosing your smartphone? (select up to 2

answers).

3) What do you most often use your

smartphone for? (select up to two answers).

Falkowski-Gilski, P. and Pankowski, M.

Energy Efﬁciency Study of Audio-video Content Consumption on Selected Android Mobile Terminals.

DOI: 10.5220/0010726500003058

In Proceedings of the 17th International Conference on Web Information Systems and Technologies (WEBIST 2021), pages 647-657

ISBN: 978-989-758-536-4; ISSN: 2184-3252

647

Figure 1: Age distribution of individuals participating in the survey.

The first question checked how “computerized”

or “mobile” was each individual. Nowadays, a typical

user has more than 1 mobile device. It is also

interesting to learn how many users use a greater

number of mobile terminals on a daily basis.

The second question allowed us to notice what

features are most desirable among users. Thanks to

this we could note their preferences. What makes a

successful mobile device and what producers should

pay special attention to.

The third question helped to determine which

tasks smartphones are often used for. This part

enabled us to design appropriate scenarios that could

reflect realistic, everyday usage.

The fourth question pointed out which aspects and

user expectations have not yet been met or fulfilled,

as well as what characteristics of the device

deteriorate over time.

The survey was posted on a social group called

“Telefonawka” on Facebook, which associates over

eighteen thousand people. It brings together users of

mobile terminals, including different software and

operating systems, IoT (Internet of Things)

accessories, as well as manufacturers and other

interested third parties. As a result, more than five

hundred people replied. The age distribution of

participants is shown in Figure 1.

The age of active participants ranges from 13 to

43 years of age. Most of the users participating in the

survey are between 15 and 21 years old. Results of

the survey, describing provided answers to each of the

3 questions, are shown in Figures 2-4, respectively.

Figure 2: Number of mobile devices used everyday.

In the surveyed group, over a half of individuals

use just one mobile device on a daily basis. Two

devices are systematically used by approx. one third

of them, whereas 14% of them use 3 and more devices

every day.

Figure 3: Factors taken into account when choosing a

mobile device.

10 15 20 25 30 35 40 45

No. of individuals

Age [years old]

56%

30%

10%

1 2 3 4 and more

14%

23%

30%

16%

15%

Appearance Price Efficiency

Battery life Camera quality Other

QQSS 2021 - Special Session on Quality of Service and Quality of Experience in Systems and Services

648

For most people (30%), efficiency (performance)

is the most important factor. The second place is

price, as many of them take into account the price-

quality radio when choosing a smartphone, etc.

Battery life came in third place, as it allows to enjoy

the aforementioned performance. Whereas, the

quality of photos and videos and the appearance of a

device (dimensions, screen size, weight, etc.) were

slightly less important. Other parameters included,

among others, the operating system, support for

updates, as well as waterproof and dustproof features.

Figure 4: Main activities with a mobile device.

When it comes to the main activities, the most

frequently chosen answer were social media.

It should be noted that this survey itself was realized

with the aid of a social media platform. The second

and third place was playback of multimedia (audio

and mixed audio-video). Whereas activities related

with voice calls and text messaging came next.

Surprisingly, only a small percentage of responders

declared their interest in games. Undeniably, console

or computer games are still more popular.

Other activities included the ability to take pictures,

record movies and navigation purposes. A related

study, carried out in Portugal, may be found in

(Horta et al., 2016).

3 ENERGY EFFICIENCY

There are many mobile device available on the

market and the number of smartphones launched each

year continues to grow. From a user’s perspective,

it is highly desirable to own a device that is both

powerful and resource efficient (Ferroni et al., 2014).

As our survey shown, battery lifetime is one of those

parameters particularly important to a wide group of

recipients.

The battery capacity itself is limited, due to the

size and weight of a portable device (Abdelmotalib

and Wu, 2012). Today’s smartphones have a lot of

different functions and applications. In order to

understand which factors can affect the energy

efficiency of a mobile device, one must determine

which parts of the operating system or daily routine

consume the largest amount of energy under different

circumstances.

A regular cell phone that does not use smart

applications can operate for several days on a single

charge. With modern smartphones, handling multiple

applications in the background, the device needs to be

charger every one or two days (Segata, Bloessl,

Sommer and Dressler, 2014). It is worth mentioning

that smartphones consume more energy compared to

regular cell phones even if their smart applications are

rarely used. While the smartphone’s battery capacity

has increased, battery lifetime proved to be shorter

compared to regular cell phones (Kim, Yun, Lee and

Choi, 2012).

Additionally, graphical capabilities of

smartphones in the last few years have grown

significantly. The progress was possible thanks to the

development of GPU (Graphical Processing Unit)

chipsets. More powerful GPUs increase battery

consumption. Today, smartphones have large, high

resolutions screens that enable to process and present

more demanding graphic data.

Of course other factors can affect battery lifetime,

such as: build-in sensors, enabled wireless modules,

services (applications) running in the background,

brightness of the screen, and of course type of

operating cellular standard or data transfer

technology (Perrucci, Fitzek and Widmer, 2011;

König, Memon and David, 2013; Schlichting and

Sawin, 2017).

4 TESTED MOBILE DEVICES

There are many mobile device available on the

market and the number of smartphones launched each

year continues to grow. From a user’s perspective,

it is highly desirable to own a device that is both

powerful and resource efficient. As the survey shown,

battery lifetime is one of parameter particularly

important to a wide group of recipients. The technical

specification of 4 tested mobile devices is described

in Table 1.

These smartphones come from various different

manufacturers, they differ in both hardware and

software parameters. The oldest of them (Smartphone

3) comes from 2012, and was a flagship model at that

time. Smartphone 4, from 2016, is an unusual model,

not intended for the European market. Smartphone 2

was manufactured in 2017, it is a mid-range phone,

39%

12%

25%

16%

Social media Watching videos

Listening to music Voice/text message

Games Other

Energy Efﬁciency Study of Audio-video Content Consumption on Selected Android Mobile Terminals

649

Table 1: Technical specification of tested mobile devices.

Smartphone 1

(Alcatel 3)

Smartphone 2

(Xiaomi Redmi 4X)

Smartphone 3

(Samsung Galaxy S3)

Smartphone 4

(Freetel Musashi)

Network

GSM: 850 900

1800 1900

GSM: 850 900

1800 1900

GSM: 850 900

1800 1900

GSM: 850 900

1800 1900

UMTS: 850 900

1900 2100

UMTS: 850 900

1900 2100

UMTS: 850 900

1900 2100

UMTS: 800 900

2100

LTE: 800 900

1800 2100 2600

LTE: 850 1800

1900 2100 2300

2500 2600

LTE: 800 900

1800 2100

Screen

Type IPS TFT IPS LCD Super AMOLED TFT

Dimensions 5.5’’ 5.5’’ 4.8’’ 4.0’’

Resolution 740 x 1440 1080 x 1920 720 x 1280 480 x 800

Battery Capacity 3000 mAh 4100 mAh 2100 mAh 2000 mAh

Platform

Operating

system

Android 8.0 Android 4.4 Android 4.0 Android 5.1

CPU

MediaTek 6739

1.28 GHz

4 Cores

MediaTek 6797

2.30 GHz

10 Cores

Exynos 4412

1.40 GHz

4 Cores

MediaTek 6735M

1.0 GHz

4 Cores

GPU

PowerVR GE8100

570 MHz

Adreno 506

650 MHz

Mali-400

440 MHz

Mali T-720

650 MHz

RAM 2 GB 4 GB 1 GB 1 GB

very popular among consumers. Whereas,

Smartphone 1 was produced in 2018, as a low-price

model, with dual-SIM capabilities.

The initial setup as well as testing procedure is

described in Table 2. The utilized benchmark

application was Trepn Power Profiler from

Qualcomm (Qualcomm, 2021), a power and

performance profiling application, designed to

identify applications that are CPU-intensive, data

consuming, or simply drain the battery. This

application can be run on any Android-powered

device with version 4.0 or higher. The app has two

modes of operation: preset and advanced mode.

The predefined preset mode enables to monitor:

1) CPU speed – displays the speed of CPU

cores on the screen.

2) Mobile data – detect which applications are

using cellular/Wi-Fi data.

3) Performance – a plot of CPU and GPU load.

4) CPU usage – generates the percentage of

usage.

5) CPU load – a plot of CPU cores load.

6) Network activity – including the status

related to the operating networks.

The advanced mode allows to select a set of

parameters that one is interested in. Of course the

availability of respective data depends on the

manufacturer of the chipset, which sometimes may

not be available. For the purpose of this study,

we have selected 2 parameters, namely: battery power

[mW], CPU load [%]. Those factors were monitored

during the use of both cellular and Wi-Fi data

transfer, as well as lowest and highest brightness

settings.

Table 2: Initial setup and testing procedure.

Stage Step Description

Initial

setup

Step 1

Device is fully charged and

powered on

Step 2

Screen brightness is set to

lowest/highest level

Step 3

Cellular/Wi-Fi data

transmission is enabled

Step 4

Testing, custom and evaluated

application are launched

Testing

procedure

Step 1

Launching all applications and

configurations

Step 2

Audio/Video playback over a

period of 3 minutes

Step 3

Ending all actions, saving

results to .csv file

Step 4 End of procedure

The execution of each step, as described, was

automated by our custom-build software, which was

later used to gather and handle obtained data.

QQSS 2021 - Special Session on Quality of Service and Quality of Experience in Systems and Services

650

5 RESULTS

Since watching videos and listening to audio proved

to be one of the most common activities with a

smartphone, we have performed a study concerning:

battery usage and CPU load. Our scenarios included

two types of data transmission (cellular and Wi-Fi),

as well as different screen brightness level (minimum

and maximum). Results of this study are shown in

Figures 5-16. Those related with consumption of

video content, particularly YouTube application,

are shown in Figures 5-12, whereas those focused on

audio content, namely Spotify application, are shown

in Figures 13-16, respectively.

Figure 5: Battery power consumption while watching YouTube videos with cellular data at lowest brightness level:

instantaneous value (left), averaged value (right).

Figure 6: Battery power consumption while watching YouTube videos with cellular data at highest brightness level:

instantaneous value (left), averaged value (right).

Energy Efﬁciency Study of Audio-video Content Consumption on Selected Android Mobile Terminals

651

Figure 7: CPU load while watching YouTube videos with cellular data at lowest brightness level: instantaneous value (left),

averaged value (right).

Figure 8: CPU load while watching YouTube videos with cellular data at highest brightness level: instantaneous value (left),

averaged value (right).

When watching movies using the mobile Internet,

the brightness of the screen is of great importance,

especially among the older models. Alcatel 3

(Smartphone 1) obtained similar power consumption

in both variants at approx. 350-380 mW, whereas

other devices recorded a higher difference of approx.

150 mW (Samsung Galaxy S3), 400 mw (Xiaomi

Redmi X4) and 500 mW (Freetel Musashi).

What is interesting, CPU load did not vary

depending on the brightness level of the screen.

QQSS 2021 - Special Session on Quality of Service and Quality of Experience in Systems and Services

652

Figure 9: Battery power consumption while watching YouTube videos with Wi-Fi data at lowest brightness level:

instantaneous value (left), averaged value (right).

Figure 10: Battery power consumption while watching YouTube videos with Wi-Fi data at highest brightness level:

instantaneous value (left), averaged value (right).

Comparing obtained results it can be seen that in

the case of Redmi Note 4X and Freetel Musashi with

brighter screen, power consumption increased

drastically (by 500 mW and 400 mW, respectively),

as did when using mobile Internet. Alcatel 3 and

Samsung Galaxy S3, on the other hand, achieved

better results with a higher screen brightness

(reduction of consumption by 120 and 130 mW).

Energy Efﬁciency Study of Audio-video Content Consumption on Selected Android Mobile Terminals

653

Figure 11: CPU load while watching YouTube videos with Wi-Fi data at lowest brightness level: instantaneous value (left),

averaged value (right).

Figure 12: CPU load while watching YouTube videos with Wi-Fi data at highest brightness level: instantaneous value (left),

averaged value (right).

As shown, the brightness of the screen does not

affect CPU usage. The load is at a similar level as in

the analogous test when using mobile Internet.

During this scenario, on the two latest phones

(Alcatel 3 and Xiaomi Redmi X4), one can notice that

the use of mobile data required much more battery

power (470 mW compared to 355 mW in case of

Wi-Fi). The difference between cellular and Wi-Fi

data transmission in case of Samsung Galaxy S3 and

Freetel Musashi was lower, yet it favored the Wi-Fi

connection as well (27 compared to 124 mW).

QQSS 2021 - Special Session on Quality of Service and Quality of Experience in Systems and Services

654

Figure 13: Battery power consumption while listening Spotify audio with cellular data: instantaneous value (left), averaged

value (right).

Figure 14: Battery power consumption while listening Spotify audio with Wi-Fi data: instantaneous value (left), averaged

value (right).

The measured CPU loads partially coincide with

the battery consumption for individual devices

(Freetel Musashi and Samsung Galaxy S3). During

the transfer via Wi-Fi they are characterized by higher

load values compared to cellular data transmission.

Automatically, this translates into higher battery

usage.

6 SUMMARY

In the carried out study, the tested devices were

subjected to typical user scenarios, in which their

energy efficiency has been put to the test. It has

shown that the degree to which the screen brightness

is selected, the choice of data transfer technology

affects the use of the battery and its rate of discharge.

Energy Efﬁciency Study of Audio-video Content Consumption on Selected Android Mobile Terminals

655

Figure 15: CPU load while listening Spotify audio with cellular data: instantaneous value (left), averaged value (right).

Figure 16: CPU load while listening Spotify audio with Wi-Fi data: instantaneous value (left), averaged value (right).

Current mobile communication includes

transferring and handling huge amounts of data

containing high-quality sound and image, which must

be processed and properly displayed on the device.

Hence, newer phones offer higher hardware

investments.

As shown, generally speaking mobile data

transmission (cellular connection) required much

more battery resources than Wi-Fi data transmission.

As expected, video playback proved to be more

demanding than just single audio. Whereas, screen

brightness should be also taken into account when

designing energy effective solutions. Additionally,

results clearly show that newer devices, both

considering hardware (build-in components) and

software (distribution of the operating system),

have a noticeable advantage over older devices.

However, this does not mean that there is really a

necessity to change one’s mobile device every year

(The Climate Group, 2008).

Still, the topic of energy efficiency and battery

consumption of mobile devices and related systems

and services remains open. Future studies may and

should include a broader range of user activities as

QQSS 2021 - Special Session on Quality of Service and Quality of Experience in Systems and Services

656

well as hardware and software platforms, including a

single or multiple operating systems and user devices,

not to mention network optimization methods and

algorithms. It would be also interesting to evaluate

various playback accessories, including loudspeakers

and headphones, both wired and wireless. A source of

inspiration may be found in (Coughlin and IEEE

Consumer Electronics Society Future Directions

Committee, 2014).

REFERENCES

Abdelmotalib, A., Wu, Z. (2012). Power consumption in

smartphones (hardware behaviourism). International

Journal of Computer Science Issues, 9(3), 161.

Coughlin, T., IEEE Consumer Electronics Society Future

Directions Committee. (2014). A Moore’s Law for

mobile energy: Improving upon conventional batteries

and energy sources for mobile devices. IEEE Consumer

Electronics Magazine, 4(1), 74-82.

Falkowski-Gilski, P. (2020). On the consumption of

multimedia content using mobile devices: A year to

year user case study. Archives of Acoustics, 45(2),

321-328.

Falkowski-Gilski, P., Uhl, T. (2020). Current trends in

consumption of multimedia content using online

streaming platforms: A user-centric survey. Computer

Science Review, 37, 100268.

Ferroni, M., Cazzola, A., Trovò, F., Sciuto, D.,

Santambrogio, M. D. (2014). On power and energy

consumption modeling for smart mobile devices.

In EUC’14, 12th IEEE International Conference on

Embedded and Ubiquitous Computing, 273-280.

Horta, A., Fonseca, S., Truninger, M., Nobre, N.,

Correia, A. (2016). Mobile phones, batteries and power

consumption: An analysis of social practices in

Portugal. Energy Research & Social Science, 13, 15-23.

Kim, M. W., Yun, D. G., Lee, J. M., Choi, S. G. (2012).

Battery life time extension method using selective data

reception on smartphone. In ICOIN’12, International

Conference on Information Network, 468-471.

König, I., Memon, A. Q., David, K. (2013). Energy

consumption of the sensors of smartphones. In

ISWCS’13, 10th International Symposium on Wireless

Communication Systems, 1-5.

Perrucci, G. P., Fitzek, F. H., Widmer, J. (2011). Survey on

energy consumption entities on the smartphone

platform. In VTC’11, 73rd IEEE Vehicular Technology

Conference, 1-6.

Qualcomm. (2021). Trepn Power Profiler.

https://developer.qualcomm.com/forums/software/trep

n-power-profiler [accessed: 15.07.2021].

Schlichting, M., Sawin, J. (2017). Modeling the effects of

independent components on mobile device charging

times. In FiCloud’17, 5th IEEE International

Conference on Future Internet of Things and Cloud,

142-147.

Segata, M., Bloessl, B., Sommer, C., Dressler, F. (2014).

Towards energy efficient smart phone applications:

Energy models for offloading tasks into the cloud.

In ICC’14, IEEE International Conference on

Communications, 2394-2399.

The Climate Group. (2008). Smart 2020 – Enabling the

Low Carbon Economy in the Information Age.

Energy Efﬁciency Study of Audio-video Content Consumption on Selected Android Mobile Terminals

657