The document discusses current and future trends in Android power management. It summarizes the Android power management architecture, including wake locks and power drivers. It also reviews present research on power consumption analysis and management, such as reducing power usage from connectivity features like Wi-Fi and GPS. The document surveys existing power saving Android apps and their limitations. It proposes future directions for more intelligent power management apps using monitoring and learning techniques while addressing privacy and security concerns.

1. Android Power Management:
Current and Future Trends
Soumya Kanti Datta
Research Engineer
EURECOM
France

2. Projet : « Smart 4G Tablet»
LPCIM : Laboratoire de Physique des
Interfaces et des Couches Minces

3. Contents
• Introduction
• Android power management architecture
• Present research on power consumption &
management
• Survey on power saving Android apps
• Future directions
• Privacy & security concerns
• Conclusion
18/06/2012 ETSIoT 2012, Seoul 3

4. Introduction
• Smart device features that consume power
– CPU, Wi-Fi, GPS, colorful bright display, 3G, auto
sync, Bluetooth etc.
– third party advertisements shown in free Android
apps
• Many power saving apps written
– control Wi-Fi, 2G and 3G connections, brightness
level, CPU frequency, GPS etc.
18/06/2012 ETSIoT 2012, Seoul 4

5. Contents
• Introduction
• Android power management architecture
• Present research on power consumption &
management
• Survey on power saving Android apps
• Future directions
• Privacy & security concerns
• Conclusion
18/06/2012 ETSIoT 2012, Seoul 5

6. Android Power Management
• Power driver addition to Linux kernel.
• Wake Locks
– It prevents the system from entering suspend or other
low-power states.
– Android apps are required to request CPU resources
with wake locks.
18/06/2012 ETSIoT 2012, Seoul 6

7. Android PM Architecture
18/06/2012 ETSIoT 2012, Seoul 7

8. Contents
• Introduction
• Android power management architecture
• Present research on power consumption &
management
• Survey on power saving Android apps
• Future directions
• Privacy & security concerns
• Conclusion
18/06/2012 ETSIoT 2012, Seoul 8

9. Power consumption of connectivity
features
• Reducing power dissipation in Wi-Fi, 3G/4G,
GPS
• Location Based Apps (LBA)
– Social networking, weather, traffic, ad etc.
– Adaptive location sensing framework
18/06/2012 ETSIoT 2012, Seoul 9

10. Context Aware PM
• Employs several monitoring system to extract
contexts (CABMAN)
– Location
– Battery
– Process
– Call
• Can predict the next charging opportunity
18/06/2012 ETSIoT 2012, Seoul 10

11. Power Model Generation
• PowerBooter technique & PowerTutor app is
introduced in [].
• Monitoring & recording battery consumption.
• Each hardware element has several states and
power consumption depends on the states.
• Power consumption is correlated with the
hardware elements of the smartphones.
18/06/2012 ETSIoT 2012, Seoul 11

12. Other Approaches on PM
• Analysis using smartphone usage pattern
– An Android app ‘battery logger’ is used to collect
usage pattern and send it to a server.
• Internal energy consumption of apps
• Human Battery Interaction
18/06/2012 ETSIoT 2012, Seoul 12

13. Contents
• Introduction
• Android power management architecture
• Present research on power consumption &
management
• Survey on power saving Android apps
• Future directions
• Privacy & security concerns
• Conclusion
18/06/2012 ETSIoT 2012, Seoul 13

14. Approach for power saving
Primary Approach Secondary Approach App Example
CPU frequency
scaling
Toggling other
features
Setcpu, cputuner
Control
smartphone
features
CPU frequency
optimization
Juice defencder,
extended control
and more
18/06/2012 ETSIoT 2012, Seoul 14

15. Increasing power efficiency
• Toggle control on Wi-Fi, Bluetooth, GPS, auto sync,
airplane mode, auto screen lock, USB mass storage,
screen-always-on, torch, 2G, 3G, 4G/Wimax (if present)
and mobile data (APN).
• Change brightness level of display.
• Volume and vibration control.
• Alter screen timeout value.
• Scheduling – night, weekend, peak.
• Setting Wi-Fi timeout.
• Setting dark home screen wallpaper for OLED display.
18/06/2012 ETSIoT 2012, Seoul 15

16. Root Access
Select CPU Governor
Advanced
Options
Condition Monitor
Appropriate
Profile
System Info
Battery Monitor
Priorit
y
Smartphone features
are controlled
Required for CPU
freq. scaling
Ondemand
Conservativ
e
interactive
Advanced
Monitoring
Single condition
Several conditions
Primary focus is CPU
freq. scaling
Example: setcpu
Overall Architecture for CPU frequency
scaling approaches
18/06/2012 ETSIoT 2012, Seoul 16

17. Condition
Monitor
Appropriate
Profile
System Info
Battery Monitor
Smartphone features
are controlled
Context Monitor
Decision
Location
Time
Battery
LevelPhone Idle
Triggers
1. Predefined
2. Customized by user
Toggle connectivity
Keep on some apps
CPU freq. scaling is
secondary focus
Example:
Juicedefender
Not used for any
learning
Overall Architecture for Control
Smartphone Features Approaches
18/06/2012 ETSIoT 2012, Seoul 17

18. Limitations of studied apps
• Static profiling (one customizable profile)
– Not customized for specific user
• “Controlling” – not intelligent
• Requirement for root access
– For CPU freq. scaling
• No focus on power consumption pattern (except power
tutor)
• Context information is not used for any learning purpose
• Do not focus on learning the power consumption pattern of
users.
• In-app advertising: waste of energy
18/06/2012 ETSIoT 2012, Seoul 18

19. Contents
• Introduction
• Android power management architecture
• Present research on power consumption &
management
• Survey on power saving Android apps
• Future directions
• Privacy & security concerns
• Conclusion
18/06/2012 ETSIoT 2012, Seoul 19

20. Apps for Power Management
• Monitoring subsystem
– Extracts several
information like
location, date & time,
battery discharge
behavior, apps running
[usage pattern]
• Controlling subsystem
– Control
smartphone/tablet
features based on
usage pattern
18/06/2012 ETSIoT 2012, Seoul 20
• Two types of
architectures
• Client server
• Learning engine

21. Overall System Architecture
18/06/2012 ETSIoT 2012, Seoul 21
Learning engine Client server
Monitoring subsystem
Control subsystem

22. Advantages of Client Server
Architecture
• Generation of smartphone usage pattern
• The context in which most power is spent
• Clustering similar usage pattern at the server
– Derive power saving profiles based on clusters
• Evolution of clusters
• Network usage pattern
– Useful for service providers
• But threat to privacy
18/06/2012 ETSIoT 2012, Seoul 22

23. Advantages of Learning Engine
• Intelligent learning engine
• Usage pattern and power saving profile are
generated locally
• No threat on private data
• But network usage pattern could not be
known
18/06/2012 ETSIoT 2012, Seoul 23

24. Other Avenues
• Adding a photovoltaic cell to generate
additional power
– WYSIPS has introduced this
• Adaptive display
– Shorten the display screen when battery level falls
below a predefined threshold
18/06/2012 ETSIoT 2012, Seoul 24

25. Contents
• Introduction
• Android power management architecture
• Present research on power consumption &
management
• Survey on power saving Android apps
• Future directions
• Privacy & security concerns
• Conclusion
18/06/2012 ETSIoT 2012, Seoul 25

26. Privacy Concerns
• Today’s smart devices contain user sensitive
information.
• Usage pattern is mutually exclusive.
• Collecting usage logs to interpret usage pattern
raises the issue of privacy.
• Generation of usage pattern must be privacy
aware
18/06/2012 ETSIoT 2012, Seoul 26

27. Privacy Concerns
• Raw usage data User behavior
• App stores info in a log file another app could
access it and send to other remote server
• Location information 3rd party ad and Location
privacy is compromised
• Some app with similar approach could also collect
– MAC address of Android device
– IP address
– Account credentials
– Phone number
18/06/2012 ETSIoT 2012, Seoul 27

28. Solutions
• Store monitoring info in a secure database instead of writing
into a file.
• Encrypt the collected information.
• The app randomly decides when to send the database dump
to the remote server.
• Database dump is sent over a secure connection.
• The user should be given some control over how much usage
information is to be sent to the remote server.
• The remote server must employ some privacy preserving data
mining algorithm to generate the usage patterns.
• Such patterns should never be revealed to any third party.
18/06/2012 ETSIoT 2012, Seoul 28

29. Repackaging Attack
18/06/2012 ETSIoT 2012, Seoul 29

30. Contents
• Introduction
• Android power management architecture
• Present research on power consumption &
management
• Survey on power saving Android apps
• Future directions
• Privacy & security concerns
• Conclusion
18/06/2012 ETSIoT 2012, Seoul 30

31. Conclusions
• Highlighting the need of efficient power
management for Android devices
• Different research directions are discussed.
• Survey on power saving apps
• Future direction on power management
• Privacy threats & repackaging attack
18/06/2012 ETSIoT 2012, Seoul 31

32. Thank you
18/06/2012 ETSIoT 2012, Seoul 32