Avoiding the Rush Hours: WiFi Energy Management via Traffic - - PowerPoint PPT Presentation

Avoiding the Rush Hours:

WiFi Energy Management via Traffic Isolation

Justin Manweiler Romit Roy Choudhury

ACM MobiSys 2011

WiFi

2


Time


Zzz…
 Zzz…


Between
packet
bursts,
WiFi
switches
 to
low‐power
sleep
mode


: Saving Energy through Sleep

WiFi Sleep Under Contention

3


Time


Zzz…
 Zzz…


Time


Zzz…
 Zzz…


Beacon Wakeups

4


Bad
wakeups
=
 burst
conten+on


Key
intui8on:
move
beacons,
spread
 apart
traffic,
let
clients
sleep
faster


Traffic

 Download


MEASUREMENTS


Energy
performance
on
modern
WiFi
smartphones


5


Zzz…
 Zzz…


vs


6


Simultaneous
measurements
at
5K
hertz


Energy Profile of Nexus One

7


With
conten*on:
 ↑
Idle/Overhear,
↓
Sleep


100 200 300 400 500 600 700 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Power (mW) Time (s)

Time (s)

Idle/Overhear
 Light
Sleep


Energy Cost of Contention

8


0
 5
 10
 15
 20
 25
 30
 35
 40
 1
AP
 2
AP
 3
AP
 4
AP
 5
AP
 6
AP
 7
AP
 8
AP
 Total
Energy
in
Joules
(J)
 Iperf
 YouTube
 Energy
costs
grow
with
 number
of
contenders


Denser
Neighborhood


File
Download


Activity Percentages

9


0%
 20%
 40%
 60%
 80%
 100%
 1
AP
 2
AP
 3
AP
 4
AP
 5
AP
 6
AP
 7
AP
 8
AP
 File
Download
 High
Power
 Idle/Overhear
 Light
Sleep
 Deep
Sleep
 Increasing
Ume
in

 Idle/Overhear


Transmit/Receive


Time


SLEEPWELL
DESIGN


Avoiding
the
rush
hours
to
save
energy


10


Wakeup
later
/
go
home
later


Smarter
commute
=
save
gas
 Smarter
beacons
=
save
ba\ery


SleepWell Techniques

• Traffic Monitoring
• APs
maintain
a
map
of
peers
in
the
wireless
vicinity


• Traffic Migration
• APs
select
a
new
beacon
posiUon
based
on
heurisUcs


• Traffic Preemption
• APs
avoid
traffic
spillover
into
that
of
neighbors


11


Traffic Monitoring

12


beacon
&
traffic
maps
for
 the
one‐hop
neighborhood


Traffic Migration

13


85
 25
 70
 55
 0
 55
 75
 50


Expected
share
=

 100/(n
+
1)
=
25
ms
 Claim
expected
share
 from
largest
hole
 CONVERGES


Traffic Preemption

14


Traffic
preemp+on
 prevents
spillover


25
 0
 75
 50


Key Implementation Challenge

• APs need to change the beacon timings
• But, no 802.11 protocol support
• Fortunately, clients synchronize to AP clocks
• AP can change beacon by “lying” about the time

15


Fully
802.11
compa+ble
AP:


Hostapd
+
modified
Atheros
Ath9k
802.11n
driver
 40


Rescheduling Client Wakeups

16


“hey
client


this
beacon
is


60ms
Late”


Actual
 Time
 Client
Clock


(sync
to
AP)


I’ll
adjust

 my
clock


OK,
I
need
to
 wakeup
in
40ms


50
 50
 0
 0


Right



• n
+me


I
know
client
will



wakeup
in
40ms
 Yes,
delayed

 client
by
40ms


0
 200
 400
 600
 800
 0.0
 0.4
 0.8
 1.2
 1.6
 2.0
 2.4
 2.8
 3.2
 3.6
 Power
(mW)
 Time
(s)
 SleepWell,
2
AP
(Client
A)
 0
 200
 400
 600
 800
 0.0
 0.4
 0.8
 1.2
 1.6
 2.0
 2.4
 2.8
 3.2
 3.6
 Power
(mW)
 Time
(s)
 SleepWell,
2
AP
(Client
B)


Energy TDMA

17


0
 200
 400
 600
 800
 0.0
 0.4
 0.8
 1.2
 1.6
 2.0
 2.4
 2.8
 3.2
 3.6
 Power
(mW)
 Time
(s)
 802.11,
2
AP


Energy Comparison

18


0
 5
 10
 15
 20
 25
 30
 35
 40
 Iperf
 YouTube
 Pandora
 Total
Energy
in
Joules
(J)
 No
ContenUon
 802.11,
8
AP
 SleepWell,
8
AP
 File
Download


Activity Percentages: 802.11

19


0%
 20%
 40%
 60%
 80%
 100%
 1
AP
 2
AP
 3
AP
 4
AP
 5
AP
 6
AP
 7
AP
 8
AP
 File
Download
 High
Power
 Idle/Overhear
 Light
Sleep
 Deep
Sleep


Transmit/Receive


Activity Percentages: SleepWell

20


0%
 20%
 40%
 60%
 80%
 100%
 1
AP
 2
AP
 3
AP
 4
AP
 5
AP
 6
AP
 7
AP
 8
AP
 File
Download
 High
Power
 Idle/Overhear
 Light
Sleep
 Deep
Sleep


Transmit/Receive


Youtube CDF, Instantaneous Power

21


0
 0.2
 0.4
 0.6
 0.8
 1
 0
 100
 200
 300
 400
 500
 600
 Empirical
CDF
 Power
in
Milliwa\s
(mW)
 1
AP
 802.11,
8
AP
 SleepWell,
8
AP


SleepWell
closely
matches
 zero‐conten8on
energy
profile


Throughput under SleepWell

(per-link TCP on 4 AP testbed)

22


0
 0.2
 0.4
 0.6
 0.8
 1
 0
 0.5
 1
 1.5
 2
 2.5
 Empirical
CDF
 Bandwidth
(Mbps)
 802.11
 SleepWell


Negligible
performance
impact:
 SleepWell
just
reorders
traffic


Limitations

• Not immediately suitable to interactive traffic (VoIP)
• True
of
802.11
PSM
in
general

• Legacy APs lessen energy savings
• Won’t
preempt
for
SleepWell
traffic

• Contention from clients of the same AP
• Considered
in
NAPman
[MobiSys
2010]


23


Prior Work

• WiFi PSM Sleep Optimization
• NAPman,
Catnap
[MobiSys
10]

• μPM
[MobiSys
08]

• WiFi Duty Cycling
• Wake‐on‐Wireless
[MobiCom
02]
/
revisited
[MobiSys
07]

• Context‐for‐Wireless
[MobiSys
07]

• Blue‐Fi
[MobiSys
09],
Breadcrumbs
[MobiCom
08]

• Also,
Turducken,
Coolspots,
Tailender,
etc.

• Sensor network TDMA
• Z‐MAC
[SenSys
05]

• S‐MAC
[INFOCOM
02]


24


Conclusion

• PSM is a valuable energy-saving optimization
• But, PSM designed with a single AP in mind
• Multiple APs induce contention, waste energy
• Staggered wakeups  clients sleep through contention
• SleepWell = PSM made efficient for high-density networks

25


Zzz…
 Zzz…


to


THANK
YOU!


Ques8ons?


26


cs.duke.edu/~jgm
 jgm@cs.duke.edu