15-446 Networked Systems
Practicum
Lecture 7 – Power Management
1
Outline
• 802.11 details
• BSD
• Catnap
• Secondary Radio Systems
• Localization
2
Power Management
Approach(Infrastructure)
• Allow idle station to go to sleep
• stations power save mode stored in AP
• AP buffers packets for sleeping nodes
• AP announces which station have frames buffered
• Power Saving stations wake up periodically
• listen for Beacons
• TSF assures AP and Power save stations are
synchronized
• TSF timer keeps running when stations are
sleeping
WiFi : Saving Energy through Sleep
Between packet bursts, WiFi
switches to low-power sleep
mode
Zzz…
Zzz…
Time
4
Simultaneous measurements at 5K
hertz
5
Wireless Interface Power-Saving
• AWAKE: high power consumption, even if idle
• SLEEP: low power consumption, but can’t communicate
• Basic PSM strategy: Sleep to save energy, periodically wake to
check for pending data
• PSM protocol: when to sleep and when to wake?
• A PSM-static protocol has a regular, unchanging, sleep/wake
cycle while the network is inactive (e.g. 802.11)
Measurements of Enterasys Networks RoamAbout 802.11 NIC
750mW
time
PSM on
power
power
PSM off
50mW
time
100ms
Power Management
Approach(Infrastructure)
• Broadcast/multicast frames are also buffered at AP
• these frames are sent only at DTIM
• DTIM = time when multicast frames are to be delivered
by AP, determined by AP
• this time is indicated in the Beacon frames as delivery
traffic indication map(DTIM)
• Power Saving stations wake up prior to expected DTIM
• If TIM indicates frame buffered
• station sends PS-Poll and stays awake to receive data
• else station sleeps again
Infrastructure Power
Management Operation
DTIM Interval
Beacon-Interval
Time axis
AP activity
TIM (in Beacon):
AP activity:
Busy medium:
DTIM:
Broadcast:
PS station
Poll
Buffered Frame Retrieval Process
for Two Stations
• Station 1 has a listen interval of 2 while
Station 2 has a listen interval of 3.
9/49
Outline
• 802.11 details
• BSD
• Catnap
• Secondary Radio Systems
• Localization
10
PSM-Static Impact on TCP
(initial RTTs)
PSM off
Mobile
Device
SYN
time
ACK
DATA
Access
Point
Server
PSM on
Mobile
Device
0ms
AWAKE
SLEEP
100ms
200ms
Access
Point
Server
PSM-Static Impact on TCP
(steady state)
PSM on
Mobile
Device
Time to send buffered window
Server RTT
window < BW•RTT
Network interface sleeps
window > BW•RTT
Network interface stays awake
Access
Point
Server
PSM-static Overall Impact on TCP
Measured TCP Performance
The transmission of each TCP window takes 100ms
until the window size grows to the product of the
wireless link bandwidth and the server RTT
Web Browsing is Slow
with PSM-static
• Web browsing typically consists of small TCP data
transfers
• RTTs are a critical determinant of performance
• PSM-static slows the initial RTTs to 100ms
• Slowdown is worse for fast server connections
• Many popular Internet sites have RTTs less than 30ms
(due to increasing deployment of Web CDNs, proxies,
caches, etc.)
• For a server RTT of 20ms, the average Web page
retrieval slowdown is 2.4x
PSM-static Does Not
Save Enough Energy
• Client workloads are bursty
• 99% of the total inactive time is spent in intervals
lasting longer than 1 second (see paper)
• During long idle periods, waking up to receive a
beacon every 100ms is inefficient
• Percentage of idle energy spent listening to beacons:
Enterasys RoamAbout 23% Used in our paper
ORiNOCO PC Gold
35% Based on data in:
Cisco AIR-PCM350
84% [Shih, MOBICOM 2002]
• Longer sleep times enable deeper sleep modes
• Basic tradeoff between reducing power and wakeup cost
• Current cards are optimized for 100ms sleep intervals
The PSM-static Dilemma
Compromise between performance and energy
If PSM-static is too coarse-grained, it harms
performance by delaying network data
If PSM-static is too fine-grained, it wastes
energy by waking unnecessarily
Solution: dynamically adapt to network activity to
maintain performance while minimizing energy
• Stay awake to avoid delaying very fast RTTs
• Back off (listen to fewer beacons) while idle
Bounding Slowdown with
Minimum Energy (Idealized)
Bounded Slowdown Property:
If Twait has elapsed since a request was sent, the network
interface can sleep for a duration up to Twait•p while
bounding the RTT slowdown to (1+p)
Twait
Twait•p
request
Idealized protocol:
• To minimize energy: sleep as much as possible
• To bound slowdown: wakeup to check for response data
as governed by above property
Synchronization
• Mobile device and AP should be synchronized
with a fixed beacon period (Tbp)
• May delay response by one beacon period
during first sleep interval
• To bound slowdown, initially stay awake for 1/p
beacon periods
• Round sleep intervals down to a multiple of Tbp
• Requires minimal changes to 802.11
(1/p)•Tbp Tbp
Bounded-Slowdown (BSD) Protocol
beacon period:
PSM-static:
BSD-100%:
BSD-50%:
BSD-20%:
BSD-10%:
• Parameterized BSD protocol exposes trade-off between
performance and energy
• Compared to PSM-static: awake energy increases, listen
energy decreases
Web Browsing Energy
• BSD would have large energy savings for other cards: 25% for
ORiNOCO PC Gold, and 70% for Cisco AIR-PCM350
• Sleep energy could be reduced by going into deeper sleep
during long sleep intervals
• Shorter beacon-period can reduce awake energy (see paper)
Outline
• 802.11 details
• BSD
• Catnap
• Secondary Radio Systems
• Localization
21
Using Sleep Modes
• 802.11 PSM
• Yes, but only when no application is using the
wireless interface
• Can hurt application performance, e.g. VoIP
• “Enter sleep mode if no network activity for X
amount of time”
• S3 Mode
• Cannot use it while applications are running
• Sleep modes not useful during data transfers
22
Typical Home Scenario
BW Discrepancy in typical end to end transfers
Client
40 Mbps
54 Mbps
Wireless AP
0.3ms
Idle period = 3.7ms
– too small for PSM
Server
3 Mbps
3-5 Mbps
Packet Transmission
Time = 4ms
Catnap leverages this opportunity to provide
up to 2 5x energy savings during data transfers
23
Catnap
Design
Catnap Design Overview
Client
Server
Catnap Proxy
1. Request
2. Request
5. Burst
3. Response
Data
Scheduler
HINT
Data
Data
4. Buffering
1. Decoupling of Wired and Wireless Segments
2. ADU Hint -- Length of ADU
3. Scheduler – Decides when to send data to client
All changes are on the proxy
40
24
Catnap
Scheduler
Scheduler
Catnap Proxy
Server
2. Request
1. Request
3. Response
Data
Data
HINT
Wired BW
Estimation
Client
4. Burst
Scheduler
Client 1
Client 2
Data
Wireless BW
Estimation
Extra
Storage
Key Points: Decoupling, Hint from Server, Scheduling
25
26
Outline
• 802.11 details
• BSD
• Catnap
• Secondary Radio Systems
• Localization
27
WiFi Sleep Under Contention
Zzz…
Zzz…
Time
Zzz…
Zzz…
Time
28
Beacon Wakeups
Bad wakeups =
burst
contention
Traffic
Download
Key intuition: move beacons,
spread apart traffic, let clients
sleep faster
29
Reducing Idle Power
The Problem
To receive a phone call the device and the wireless NIC has to be in a
“listening” state i.e. they have to be on.
Our Proposal
When not in use, turn the wireless NIC and the device off.
Create a separate control channel. Operate the control channel using
very low power, possibly in a different freq. band. Use this channel to
“wake-up” device when necessary.
Proof of Concept & Implementation
Short Term: Add a low power RF transceiver to the 802.11 enabled
handheld device
Long term: Integrate lower power functionality into 802.11 or integrate
lower power radio into mother board and/or 802.11 Access Points.
The MiniBrick PCB
Accelerometer
Tilt Sensor
IR Range
Speaker
915 MHz Radio
Crystal
PIC
Vibrator
Temperature
Sensor
Audio Plug
Front View
Back View
Modular design allows removal of components
Radio Power Consumption
Radio:
• RFM TR 1000 ASH
• Modulation: ASK
• Voltage: 3V
• Range: 30 feet (approx)
Comparing against 802.11 and BT Radios
Chipset
Receive
(mW)
Transmit
(mW)
Standby
(mW)
Rate
(Mbps)
Intersil PRISM
2 (802.11b)
400
1000
20
11
Silicon Wave
SiW1502 (BT)
160
140
20
1
RFM TR1000
14
36
0.015
0.115
MiniBrick Power Consumption
Mode
Power
Consumption
Transmit
39 mW
Receive
16 mW
Standby
7.8 mW
Theses numbers include the power consumption by the PIC Microcontroller
and the RFM TR1000
Outline
• 802.11 details
• BSD
• Catnap
• Secondary Radio Systems
• Localization
34