02 Yu Xiao Practical Power Modeling of Data Transmission 01

8/10/2019 02 Yu Xiao Practical Power Modeling of Data Transmission 01

1/23

Practical Power Modeling of Data

Transmission over 802.11g for WirelessApplications

Yu Xiao Petri Savolainen Arto Karppanen Matti SiekkinenAntti Yl-Jski

E-Energy 2010


2/23

Outline

Introduction

Related Work

Power Model

Model Validation

Discussion Conclusion


3/23

Introduction

Power consumption of data transmission in WLAN

802.11 Wireless network interface (WNI)

Different power consumption in different operating modes

Energy = Power(operating mode)* Duration(operating mode)

The duration information is not easily accessible

Estimate the operating modes & durations

802.11 power saving mode (PSM) Traffic burstiness


4/23

Related Work

Power analysis of network protocols

Power analysis of different TCP versions such as Reno,

Newreno and SACK[9]

Impact on power consumption from different TCP

header options such as window scale option[10]

Power consumption of MAC/PHY layer overhead[14]

Power models that use low-level information Power model based on WNI operating modes [3]


5/23

Power Model

Traffic Burstiness

Burst size SBBin rate r

r = SB/T = SB/(TB+TI)

In a Burst: {Packet interval < Threshold}

Burst

Duration TB

Burst

Interval TI

Packet

IntervalBin Duration T= TB+TI


6/23

WNI Operating Modes: CAM vs. PSM

Continuously Active Mode

(CAM)

Power Saving Mode(PSM)

IDLE

TRANSMITRECEIVE

SLEEP

PS

TRANSMIT

PT

IDLE PI

RECEIVEPR

PSMTimeout

Tsleep = TI - Ttimeout


7/23

Two Scenarios

Threshold of bin rate rc

When Tsleep

= 0,

rc = SB/ (TB+ Ttimeout).

Scenario 1: {{r>= rc } and {PSM is enabled}} or{CAM is enabled} .

Scenario 2: { r< rc } and {PSM is enabled}.


8/23

Downlink Power Consumption

TB

TI

Power(W)

Ps

PI

PR

PT

SLEEP IDLE RECEIVE TRANSMIT

Energy(J): E = PRTB+ PITI

Power(W): Pd(r) = E/T

Energy Utility (b/J): E0(r) = r/Pd(r)


9/23

Downlink Power Consumption

Scenario 1

E = PRTB+ PITI

Pd(r) = E/T = PI + r(PR PI) TB/SB

Scenario 2

E = PRTB+ PITtimeout + PSTsleep

Pd(r) = E/T =PS + r[(PR PS) TB/SB+ (PI PS) Ttimeout/SB]

Time

PR

PI

TB TB+ TI

Power

Time

PR

PI

TB TB+ TI

Power

TB+ TI+ Tsleep


10/23

TCP Power Consumption

Downlink data rate rdDownlink burst size SB

Uplink data rate ru = nSACKrd/SB

Data rate threshold rc = SB/(Td+Tu+Ttimeout)

Scenario 1:P(rd)=Pd(rd)+Pu(ru)-PI= PI+[Td(PR-PI)+Tu(PT-PI)]rd/SB

Secenario 2:P(rd)=Pd(rd)+Pu(ru)-PS

=PS+[Td(PR-PS)+Tu(PT-PS)+Ttimeout(PI-PS)]rd/SB

n ACKs

n packets

Td Tu


11/23

Simplified TCP Power Model

Drop the power consumption caused by ACKs

Scenario 1:

P(rd) = PI + (PR-PI)rd/rmax

Total energy consumption of receiving m bins:

Scenario 2:

P(rd) = PI + (PR-PS)rd/rmax+ (PI-PS)rdTtimeout/SB


12/23

Multiple TCP Flows

The number of TCP flows: n

The aggregate TCP throughput :

The throughput of the ith flow: ri

TCP download Power consumption: Replace the rd with


13/23

Validation

Experimental Setup


14/23

Internet flow characteristics

Nokia

N810

HTC G1 Nokia N95

Downlink burst size (KB) 4 4 4

Downlink burst duration (ms) 8 10 10

Uplink burst duration (ms) 0.5 0.5 0.35

NokiaN810

HTC G1 NokiaN95

Uplink burst size (KB) 4 4 4

Uplink burst duration (ms) 6 8 12

Downlink burst duration (ms) 0.1 0.1 0.2

0.00.10.20.30.4

0.50.60.70.80.91.0

0 30 60 90 120 150 180 210 240 270 300

Pro

bability

Packet Interval (ms)


15/23

WNI operating mode

Power consumption in different operating mode

WNI operating mode Average Power (W)

Nokia N810 HTC G1 Nokia N95

IDLE 0.884 0.650 1.038

SLEEP 0.042 0.068 0.088

TRANSMIT 1.258 1.097 1.687

RECEIVE 1.181 0.900 1.585


16/23

Nokia N810

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 32 64 96 128 160 192 224 256

AvgP

ower(W)

Data rate limit (KB/s)

Download, CAM

Measured

Estimated0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 32 64 96 128 160 192 224 256

AvgP

ower(W)

Data rate limit(KB/s)

Download, PSM

Measured

Estimated

0.0

0.20.4

0.6

0.8

1.0

1.2

0 32 64 96 128 160 192 224 256

AvgPower(W)


Upload, CAM

Measured

Estimated 0.0

0.20.4

0.6

0.8

1.0

1.2

0 32 64 96 128 160 192 224 256

Avg

Power(W)


Upload, PSM

Measured

Estimated


17/23

HTC G1

0.0

0.2

0.4

0.6

0.8

1.0

0 32 64 96 128 160 192 224 256

AvgP

ower(W)


TCP Download

Measured Estimated

0.0

0.2

0.4

0.6

0.8

1.0

0 32 64 96 128 160 192 224 256

AvgP

ower(W)


TCP Upload

Measured

Estimated


18/23

Nokia N95

0.00.20.40.60.8

1.01.21.41.6

0 32 64 96 128 160 192 224 256

AvgPo

wer(W)


Download, CAM

Measured

Estimated0.00.20.40.6

0.81.01.21.41.6

0 32 64 96 128 160 192 224 256

AvgP

ower(W)


Download, PSM

MeasuredEstimatedRefined

0.00.20.4

0.60.81.01.21.41.61.8

0 32 64 96 128 160 192 224 256

AvgPower(W)


Upload, CAM

MeasuredEstimated

0.00.20.40.6

0.81.01.21.41.61.8

0 32 64 96 128 160 192 224 256

Avg

Power(W)

Data Rate Limit(KB/s)

Upload, PSM

Measured

Estimated

Refined


19/23

Accuracy

Download cases

MAE: less than 0.068394W

MAPE: less than 6.7724%.

Upload cases

MAE: less than 0.055923W

MAPE: less than 5.7599%

0.0

0.2

0.4

0.6

0.8

1.0

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

F(MAPE)

MAPE

0.0

0.2

0.4

0.6

0.8

1.0

0 0.1 0.2 0.3 0.4 0.5

F(MAPE)

MAPE


20/23


21/23

Discussion and Future Work

Runtime Power Estimation

Network simulation

Energy-efficient network transmission


22/23

Conclusion

Usage of WNI Internet flow characteristics

(e.g. network throughput)

Traffic pattern

(e.g. Burstiness)

WNI operating mode 802.11 Power Saving Mode


23/23

Thank you! Questions?

Contact:Yu Xiao

[email protected]

POBox 15400, Aalto UniversityKonemiehentie 2, Espoo, Finland
mailto:[email protected]:[email protected]

02 Yu Xiao Practical Power Modeling of Data Transmission 01

Documents