snooze: energy management in 802.11n wlans ki-young jang, shuai hao, anmol sheth, ramesh govindan

Snooze: Energy Management in 802.11n WLANs

Ki-Young Jang, Shuai Hao, Anmol Sheth, Ramesh Govindan

Background

2

Evolution of Wi-Fi

Performance vs. Energy Usage

WaveLAN (1995)

802.11b (1997)

802.11g (2003)

802.11n 3x3

(2009)

802.11n 4x4

(2011)

0

200

400

600

800

1000

1200Wi-Fi Performance

8x speedup compared to

802.11a/g

Mbp

s

3

Background

802.11n Features and Energy Usage

A B

Energy management should exploit bothsleep opportunities and antenna configuration.

MIMO higher data rate spatial diversity

Additional power states:10-30% of peak power

consumption of the tablet!

# of Antenna

12 1.99 1.27 1.13 0.103 2.10 1.60 1.45 0.10

Intel Wi-Fi Link 5300(W)Tx Rx Idle Sleep

1.28 0.94 0.82 0.10

1.5x and 1.6x 1.3x and

1.7x

4Motivation Design Implementation Evaluation

Motivation

Micro-sleep Opportunities

A

B

APt

Sleep!Z ZZ

Sleep!Z ZZ

t

t

5

Micro-sleepSleep + Wakeup = ~2ms

Motivation Design Implementation Evaluation

Motivation

Micro-sleep Opportunities

# of STATraffic

1 1 2 1 2VoIP File HD Video

0.00.20.40.60.81.0

EXPLOITABLE NON-EXPLOITABLE TX RX

Nor

mal

ized

Tim

eSleeping while AP is servicing others1Sleeping during inter-frame gaps2

Depending on the traffic and number of clients, we can get energy savings of 30% ~ 90%.

6Motivation Design Implementation Evaluation

Motivation

1 2 30

5

10

15

RX Antennas

Tim

e (s

econ

ds)

1 2 305

101520

RX Antennas

Ener

gy (J

)

Internet as bottleneck: SISO is best

1 2 302468

10

RX Antennas

Tim

e (s

econ

ds)

1 2 302468

10

RX Antennas

Ener

gy (J

)

High bandwidth scenario: MIMO3 is best

Antenna Configuration

Antenna configuration should be adaptive based on traffic demand and link quality.

7

Challenges

• Micro-sleep with minimal impact on delay and throughput-sensitive traffic

• Adaptive antenna configuration management• Joint design of both mechanisms• Application agnostic

Time (ms)

Pow

er(W

)

100 200 300

Sleep for 50ms

Sleep for 70ms

8

Contribution

Design and Implementation of Snooze

Joint, application-agnostic design of client micro-sleep and antenna configuration management.

Extensive experiments that demonstrate 30~85% energy-savings over CAM across a wide range of traffic scenarios.

9Motivation Design Implementation Evaluation

AP-Directed Design

Snooze AP

Snooze Client

• Shapes traffic to create sleep opportunities• Minimal impact on traffic• Minimizes the number of active

clients

• Manages antenna configurations• Minimizes antennas needed

Goal: Reduce client energy consumption by jointly controlling sleep and antenna configuration

10

Sleep duration: based on measured packet arrival rate Awake duration: based on average airtime consumption

Motivation Design Implementation Evaluation

Snooze Components

Micro-sleepScheduling

AntennaMgmt.

If measured airtime utilization is

< 0.3: high link quality or more antennas -> turn off 1 antenna

> 0.7: low link quality or less antennas -> turn on 1 antenna

Rate Adaptation: AP uses default rate-control algorithm with restricted search space

Hysteresis and moving averages

SleepAwake

25.50 25.51 25.52 25.53 25.54 25.55Sleep

Awake

A: 1Mbps

B: 20Mbps

Time (Second)

11Motivation Design Implementation Evaluation

Implementation

AP Client

Driv

erKe

rnel

RateTable

AirtimeUsage

iwlagn

Airtime Scheduling

Sleep/WakeupComputation

mac80211...

Per client traffic queue

Sleep/Wakeup

Antenna Configuration

iwlagn

mac80211

12Motivation Design Implementation Evaluation

Evaluation

Applications used for evaluation

Delay/Jitter Sensitivity

High Low

Bandwidth Requirement

High HD video streaming File downloading

Low VoIP Chat

Overview and Setup

Performance comparisonConstantly Awake Mode (CAM)

802.11 Power Save Mode (PSM)

Evaluation metricTotal energy usage of NIC

Application throughput and delay

13Motivation Design Implementation Evaluation

Evaluation

High Definition Video Streaming

More than 25% low-power sleeping compared with both CAM and PSM

About 50% energy savings compared with both CAM and PSM

Average delayCAM : 2.5ms, PSM : 4ms, Snooze : 8ms

14Motivation Design Implementation Evaluation

Evaluation

Heterogeneous Traffic

Snooze can accommodate multiple concurrent applications.

Client

Mode

HD

Chat

VoIP File HD

Chat

VoIP File HD

Chat

VoIP File

CAM PSM Snooze

0

50

100

150

200 SLEEP IDLE TX RX

Ener

gy(J

)

One app per client

Both techniques contribute significantly to energy savings, but contribution varies across traffic.

Energy saving breakdown

37%

63%

File80%

20%

VoIP

69%

31%

HD90%

10%

Chat

micro-sleepantenna config.

15

Related Work

Energy Management Techniques for 802.11

uPM C-PSM Catnap NAPman Snooze

AP-directed

Traffic types

HB-DS

HB-DI

LB-DS

LB-DI

802.11n

Rate adaptation

Multiple apps per client

16

Conclusion

Client micro-sleep and antenna configuration managementApplication agnostic30~85% energy saving across a wide range of traffic scenarios

Snooze: Energy Management Scheme for 802.11n

Future Work

Multi-AP settingHighly bursty workloadsParameter sensitivity