Energy-Efficient Rate-Adaptive GPS-based Positioning for Smartphones

Jeongyeup Paek, Joongheon Kim, Ramesh Govindan

CENS TalkApril 30, 2010

2

Problem

• Many emerging smartphone applications require position information to provide location-based or context aware services.

• GPS is often preferred over GSM/WiFi based methods.

• But, GPS is extremely power hungry!– Can drain phone battery in few hours.

WPS

GSM

GPS

Average ErrorGPS: 23.2mWPS: 36.8mGSM: 313.6m

WPSGPS

GSM

Average ErrorGPS: 8.8mWPS: 32.44mGSM: 176.7m

0 50 100 150 200 250 300 350 400 450 5000

0.2

0.4

0.6

0.8

1 poweravg

Time (Seconds)

Pow

er (W

att)

Off (0.06) On (0.44)

3

GPS (In)accuracy

Actual Path taken

Incorrect GPS Path

Never went here

A

B

C

GPS may provide less accurate positioning in urban areas, especially for pedestrian use

Then… can we sacrifice a little accuracy in exchange for significant reduction in energy usage?

Dis

tanc

e (m

eter

)

0

50

100

150

200

250

300 Average Distance ErrorMaximum Distance Error

Samples (70 locations)

No received GPS signals

Relatively less clear view of the sky

4

Periodic Duty-Cycling• There exist uncertainty

– Key challenge is to decide T

• Accuracy vs. Energy trade-off exists

• May introduce significant uncertainty

T

Uncertainty(distance)

Energy

….

5

RAPS : Rate-Adaptive Positioning System• An energy-efficient positioning system that adaptively duty-cycle GPS

only as often as necessary to achieve required accuracy based on user mobility and environment

• Design Goal– Reduce the amount of energy spent by the positioning system while still

providing sufficiently accurate position information– Trade-off position accuracy for reduced energy

• Challenge– Determine when and when not to turn on GPS efficiently using the sensors

and information available on a smartphone

6

RAPS Components• Movement Detection

– Use duty-cycled accelerometer with onset detection algorithm to efficiently measure the activity ratio of the user

• Velocity Estimation– Use space-time history of the past user movements along with

their associated activity ratio to estimate current user velocity

• Unavailability Detection– Use celltower-RSS blacklisting to detect GPS unavailability (e.g.

indoors) and avoid turning on GPS in these places

• Position Synchronization– Use Bluetooth-based position synchronization to reduce

position uncertainty among neighboring devices

When to turn on GPS

When NOT to turn on GPS

7

Activity Detection• Use accelerometer to detect user motion

– Binary sensor to detect non-movement

– Measure activity ratio• Onset detector for identifying activity

– Duty-cycle it for energy efficiency• 5 min accelerometer consumes

more energy than 1 min GPS0 50 100 150

0.5

1

1.5

Time (sec)

Acc

ele

ratio

n (

g)

Acceleration

Activity DetectedSignal Envelope

Activity

0 50 100 150 2000

0.1

0.2

0.3

0.4poweravg

Time (seconds)

Pow

er (W

att)

Off (0.062) On (0.141)

0 10 20 30 40 500

5

10

15

20

Duty Cycle (%)

Err

or

(%)

Operating point: 12.5%

8

Velocity Estimation• Use history of user positions

– Associate average velocity and activity ratio to particular space and time– Use these information to estimate current user velocity– Using this velocity, calculate uncertainty and decide when to turn on GPS

A

BD

C

9

Celltower Data for Movement Detection?• Celltower and RSS data cannot reliably measure user movement

0 20 40 60 80 100 120 140 160 180 2000

20

40

60

80

100

Distance between cellID changeMax. distance within same Cel-lID

Distance (meters)

Cum

ulati

ve F

racti

on (%

)

0 5 10 15 20 25 300

100

200

300Max. DistanceAvg. DistanceMin. Distance

RSSI difference (dbm)

Dis

tanc

e (m

eter

)

10

Celltower-RSS Blacklisting

• However, it can detect GPS unavailability– Signatures exist for indoor places that you go often

70 75 80 85 90 95 100 105 110 115 1200

20

40

60

80

100 cell ID 1cell ID 2

WCMDA Celltower Signal Strength (-dbm)

GPS

Ava

ilabi

lity

(%)

Good BadVariable

Turn on GPS only when available!

11

Bluetooth Position Synchronization• Use Bluetooth to synchronize position

information with neighboring nodes– Cheaper than GPS– Little uncertainty

• Short communication range (~10m)

– Bluetooth is widely being used

Save energy by lowering overall uncertainty and reducing the number of GPS activations

12

Bluetooth Position Synchronization• Use Bluetooth to synchronize position

information with neighboring nodes– Cheaper than GPS– Little uncertainty

• Short communication range (~10m)

– Bluetooth is widely being used

• Saves energy by 43% in 2-node example– TX costs ~3.07 Joule, RX costs ~1.58 Joule– GPS activation for 60sec costs ~22 Joule– (22 * 1 + 3.07 + 1.58) / (22 * 2)

• More the merrier!– Energy cost is amortized over the number

of nodes in neighborhood– For 5 nodes, 74% reduction in energy

0 50 100 150 2000

0.2

0.4

0.6

0.8

1

poweravg

Time (seconds)Po

wer

(Watt

)

0 50 100 150 2000

0.2

0.4

0.6

0.8

1

poweravg

Time (seconds)

Pow

er (W

att)

Off

Off

Bluetooth Master

Bluetooth Slave

On

On Listen & connect

Device Discovery

RX

TX

Listen

Listen

Off

Off

13

Evaluation

• Benefits of RAPS– Energy savings achieved by RAPS – Contribution of individual components

• Comparison to periodic GPS strategy

• Flexibility – Integration with WPS

• Pervasiveness of GPS errors– GPS vs. AGPS– Different platforms

14

Benefits of RAPS

LabClass

LibraryLunch

Shopping

Home

USC

~3 miles

0.4 miles

Class

History Accel.Cell-RSS Blacklist

BPS

RAPS (2)

RAPS-B

RAPS-BC

RAPS-BCA

Always-On Periodic GPS with 20 seconds interval

• Energy savings achieved by RAPS • Contribution of individual components

• Methodology– 6 phones with 5 different schemes– around USC campus area (in & outside buildings)– 34 hours

15

Benefits of RAPS - Lifetime• RAPS’s lifetime is 3.87 times longer than that of Always-On

– Each of its components contribute to this saving

RAPS RAPS-B RAPS-BC RAPS-BCA Always-On (20sec)

0

5

10

15

20

25

30

35

4034:41

31:53

16:42 16:19

8:57Life

time

(hou

rs)

Tested Schemes

3.87 times longer lifetime!

BSP – 10.8%

Blacklist – 59.0%

Accel – 1.5%

History – 28.5%

16

Benefits of RAPS - Reasons

Avg. GPS Activation Interval Expected Avg. Power Usage

(20)

GPS

Inte

rval

(sec

onds

)

Tested Schemes

RAPS RAPS-B RAPS-BC RAPS-BCA Always-On (20sec)

0

100

200

300

400

500

600

700630.9

588.5

259.5

135.4

Estim

ated

Ave

rage

Pow

er (W

)Tested Schemes

RAPS RAPS-B RAPS-BC RAPS-BCA Always-On (20sec)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4 BluetoothAccelerometerGPS

17

Did BPS work?• Contributed 10.8% of the total RAPS lifetime savings

– Defers GPS activation– Bi-directional natural incentive for sharing

BPS EnabledNode 2

BPS EnabledNode 1

GPS Activation:

BPS Communication:

BPS Disabled

18

• Contributed 59% of the total lifetime increase– Significantly increase the average interval between GPS activations– For majority of cell towers, GPS position fixing never fails– For a smaller number of cell towers, mostly those that the user visits often,

GPS failures do occur

Did Celltower-RSS Blacklist work?

0

20

40

60

80

100

0

20

40

60

80

100

120

140

GPS failGPS successsuccess ratio

Observed Cell-towers

Succ

ess

Ratio

(%)

Observation Count

Do not turn on GPS when not available!

19

Comparison to Periodic GPS• RAPS consumes,

– 48% less power compared to periodic GPS with comparable average uncertainty– 22% less power compared to periodic GPS with comparable success ratio

30 90 150 210 270 330 390 450 510 5700

20

40

60

80

100

120

140

160

180

Dis

tanc

e (m

eter

)

Periodic Interval (Seconds)

60 120 180 240 300 360 420 480 540 6000

10

20

30

40

50

60

70

80

90

100

Periodic Interval (Seconds)

Succ

ess

Ratio

(%)

85.8 m

Average Distanceachieved by RAPS

72.2%

Success Ratioachieved by RAPS

1.47 times longer lifetime

1.14 times longer lifetime

20

Is RAPS flexible?• Integration with WPS, a WiFi Positioning System

– Energy cost include WiFi scanning and data communication with database server

– Known to be less accurate than GPS

• Yes!!– Consumes less energy

• Faster position fix and turn off time

– Lower accuracy

Pow

er (W

att)

Time (Seconds)0 20 40 60 80 100 120 140

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8poweravgAvg.

Positioning Interval

Avg. Power Usage

Avg. Uncertainty

RAPS - GPS 465.1 sec 0.064 W 85.8 m

RAPS - WPS 387.3 sec 0.035 W 122.9 m

21

Are GPS Errors Pervasive?

AGPS vs. GPS Four different types of phones

0

20

40

60

80

100

120

140

160

180 GPS with G1 (avg. error: 38.4m)A-GPS with G1 (avg. error: 36.1m)GPS with N95 (avg. error: 20.2m)A-GPS with N95 (avg. error: 19.1m)

Samples (35 Locations)

Dis

tanc

e (m

eter

)

0

20

40

60

80

100

120

140

160

180 G1 (avg. error: 36.1m)N95 (avg. error: 19.1m)MotoDroid (avg. error: 18.2m)WinMobile HTC (avg. error: 13.7m)

Samples (35 Locations)D

ista

nce

(met

er)

Relatively less clear view of the skyRelatively less clear view of the sky

Similar(except G1)

Negligible difference(difference only in FTTF)

22

Conclusion and Future Work

• RAPS is a rate-adaptive positioning system for smartphone applications– GPS is generally less accurate in urban areas, so it suffices to turn on GPS

only as often as necessary to achieve this accuracy– Uses collection of techniques to cleverly determine when to and when

not to turn on GPS– Increases lifetime by factor of 3.8 relative to Always-On GPS

• Future Work– Parameter settings (e.g. accelerometer duty cycling)– Privacy and security for position sharing– User study (variety, consistency, handling behavior, etc.)

Thank you