sensing on smartphones

idleheadedceleryΚινητά – Ασύρματες Τεχνολογίες

10 Δεκ 2013 (πριν από 3 χρόνια και 6 μήνες)

49 εμφανίσεις

Improving
energy efficiency
of
location
sensing on smartphones

이시혁


theshy@sclab.yonsei.ac.kr

Z.
Zhuang

et al.,
in
Proc.
of ACM
MobiSys

2010
, pp. 315
-
330, 2010.

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Contents


Introduction


Motivation


Energy consumption of GPS


Energy
consumption of
GPS and NET


Problem characterization


Design principles


Sensing Substitution(SS)


Sensing suppRession(SR)


Sensing Piggybacking(SP)


Sensing Adaptation(SA)


Integrated operation


Software architecture and system implement


Performance evaluation


Conclusion



1

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Motivation


Location
-
based applications(LBAs) have become increasingly
popular




Most smartphones have two location sensing mechanisms


GPS(Global Positioning System)


NET(Network
-
based Triangulation : reachable cell tower, Wi
-
Fi AP)








2

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Energy consumption of GPS


3

Drop to 94% : GPS disabled

Drop to
79%
: GPS
enabled

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Energy consumption of GPS and NET

4

Drop to
93%
:
NET only

Drop to
83%
:
GPS only

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Problem characterization


Static use of location sensing
mechanisms



Absence of use of power
-
efficient sensors to optimize
location
sensing



Lack of cooperation among multiple
LBAs



Unawareness of battery level

5

GPS

NET

ACC

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Sensing Substitution(SS)


Tradeoff in Power, accuracy, and availability


Static selection


Assume GPS is always better than NET




6

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Sensing
suppRession
(SR)


Continuous sensing might be wasteful



Use of low
-
power sensor for state detection



False positive or negative on movement



7

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Sensing Piggybacking(SP)


Multiple LBAs cause duplicate GPS sensing



One
-
time registration can be monitored



Multi
-
time registration matters

8

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Sensing Adaptation(SA)


Users might prefer longer operating time



Adjust
sensing parameters (time, distance)



Adaptation
degree (e.g., 200%:
30s
-
> 1min
)


9

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Integrated operation


SS : GPS or NET


SR : Using other sensor(
low
-
power sensor
)


SP : Duplicate
GPS
sensing(multiple LBAs)


SA :
Adjust sensing parameters


10

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Software architecture and system implementation


Android framework


Middleware solution



Implement overview


OS version 1.5 Cupcake


Inside the default “Security & location”



11

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Sensing Substitution (SS)


Area 1, both
GPS

and
Net

are available


accuracy:
GPS>Net



Area 2, both

GPS
and
Net

are available


accuracy:
GPS = Net (similar)



Area
3, only GPS is available



Area
4, only Net is available


12

Performance evaluation

S FT COMPUTING @ YONSEI UNIV . KOREA

16


Static and moving



Using accelerometer

13

Sensing
SuppRession

(SR)

Performance evaluation

S FT COMPUTING @ YONSEI UNIV . KOREA

16


Using two LBAs with different starting time


14

Sensing Piggybacking (SP)

Performance evaluation

S FT COMPUTING @ YONSEI UNIV . KOREA

16


Two scenarios with different battery levels



Sensing updates every 1minute
-
> every 2 minutes


15

Sensing Adaptation (SA)

Performance evaluation

S FT COMPUTING @ YONSEI UNIV . KOREA

16


Trace from particular user


C
ommuter


Route in Silicon Valley, California



Setup


Running two LBAs concurrently at low battery level (SA, SP)


Adaptation degree : 200%


Same GPS sensing frequency of every 30 seconds


Starting with 15
-
second difference




16

Integrated results

Performance evaluation

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Conclusion


We
consider the problem of
energy efficient

location sensing on
smartphones



Four
critical factors


static
use of location
sensing mechanisms
(SS)


absence
of use of power
-
efficient sensors to
optimize location
sensing
(SR)


lack
of sensing cooperation among
multiple LBAs
(SP)


unawareness
of battery
level
(SA)



Prototype


Android OS : modified the
application framework


Reduce the GPS usage by
up to 98
%

(static state)


Improve
battery life
by
up to 75%

17

S FT COMPUTING @ YONSEI UNIV . KOREA

16

Performance evaluation


Sensing
Substitution




Sensing Suppression




Sensing
Piggybacking




Sensing
Adaptation

18

Appendix