Wake-on-WLAN Nilesh Mishra, Kameswari Chebrolu, Bhaskaran Raman and Abhinav Pathak IIT Kanpur Wake-on-WLAN Nilesh Mishra, Bhaskaran Raman, Abhinav Pathak.

Wake-on-WLANWake-on-WLAN

Nilesh Mishra, Kameswari Chebrolu,Bhaskaran Raman and Abhinav Pathak IIT Kanpur

Wake-on-WLAN

Nilesh Mishra, Bhaskaran Raman, Abhinav PathakDepartment of Computer Scienceand Engineering, IIT Kanpur

Power management for 802.11 mesh networks using 802.15.4

Kameswari ChebroluDepartment of Electrical Engineering, IIT Kanpur

Wake-on-WLANWake-on-WLANNilesh Mishra, Kameswari Chebrolu,Bhaskaran Raman and Abhinav Pathak IIT Kanpur

802.11 Mesh Network

• 802.11 designed for indoor usage

• Usage of 802.11 for long distance connectivity

• Power as a constraint

• Lack of support in current hardware

• Wake-on-WLAN

• Introduction• Related Work• Motivation• Results

• Prototype • Recent Addition• Results • Conclusions

Summary: Current 802.11 hardware is not power efficient but is being used to provide long distance connectivity


Power Consumption Measurements

• A typical mesh node– Single board

computer based 802.11 bridge/router

– Directional Antenna– RF cable for

connections

*http://www.hyperlinktech.com/web/hg2424g.php

*




Power Consumption Measurements (continued)

Observations on power consumption:• Increases with insertion of card• Higher for Tx than Rx• Considerable variation over different card makes

Summary: Even idle power consumption is significant




Trivial solution

Node 1Node 1

Node 2Node 2

Node 3Node 3

Node 4Node 4

Wired GatewayWired Gateway




Requirements

Node 1Node 1

Node 2Node 2

Node 3Node 3

Node 4Node 4





WOW!

Node 1Node 1

Node 2Node 2

Node 3Node 3

Node 4Node 4





IEEE 802.15.4

• Energy optimized

• Low cost radio (< $5)

• Works in the same 2.4GHz as 802.11

• Sensor motes

MicrocontrollerMicrocontroller

Power Supply Subsystem

Power Supply Subsystem

Communication Subsystem

Communication Subsystem

Sensing SubsystemSensing Subsystem

Application Layer Subsystem

Application Layer Subsystem

CPU

Memory

RadioRadio

Software

Sensor

ADC/DAC

BatteryBattery




Architecture

Node 1Node 1

Node 2Node 2

Node 3Node 3

Node 4Node 4


• Nodes turned off• Remote turn on• Multi-hop setting• Delay due to boot

up




Architecture

RF switchor splitter

Battery

Power switching

circuit

Node1

Antenna

Node2




CCA modes of 802.15.4• Clear if energy below

threshold• Clear if valid 802.15.4

packet • Clear if valid 802.15.4

packet and energy below threshold

• Use of Chipcon’s CC2420 CCA mode

• Configurable frequency and energy threshold parameters

• Relay based switching circuit– Does not store the state

Implementation Details• Introduction

• Related Work• Motivation• Results


Summary: 802.15.4 compliant radio is able to detect 802.11 traffic


Wake-on-WLAN Features

• On-demand, course-grained power on/off of networking equipments at a remote site.

• Use of off the shelf 802.15.4 compliant sensor motes working in 2.4 GHz.

• Setting ideal for rural deployment.

• Usage of data channel itself for remote wake-up

• No separate antenna (shared with 802.11 equipment).




• A laptop with D-Link DWL650 802.11b card running ‘trafficgen’ application as data traffic source

• Packets of size 1462 bytes at 1Mbps with inter packet intervals of 10, 20 and 100ms

• Polling of CCA pin on sensor node every 3ms

Detection of 802.11 Transmission



Summary: Traffic pattern of 802.11 successfully replicated on motes


• Validation on 3.5Km IITK-Mohanpur link on DGP testbed

• External antenna connected

• Calibrated 802.11 card for determining RxPower

Outdoor Evaluations• Introduction



Summary: Low sensitivity of 802.15.4 essential for working of Wake-on-WLAN in long distance settings


• Power consumption of Soekris acting as a router/switch with two wireless 802.11b cards = 7.7W (typical)

• Boot-up time for Soekris 50s

• VoIP service in Sarauhan in the DGP testbed– Usage pattern: 15 calls/day of 71s avg duration

• Sensor mote typical values: Vcc = 2.8V and I = 23mA with CPU and Radio receiving.

Example Scenario• Introduction




• Eno_wow = Pup x Tup

• Ewow = Emote + Ebootup + Eusage where

– Emote = Vmote x Imote x Tidle

– Ebootup = Pbootup x Tbootup

– Eusage = Pup x Tactive

• Using above values – Eno_wow = 120 Whrs

– Emote = 1.54Whrs

– Ebootup = 1.04Whrs

– Eusage = 7.73Whrs

Example Scenario

• Thus power saving is:– (Eno_wow – Ewow)/ Eno_wow

– Greater than 91%




• Does not work on 802.11a

• Hibernation facility desired: faster boot-up

• Suffers from noise generated triggers

• Advanced usage: morphing topology

• More detailed study of usage pattern for better power savings.

Discussion• Introduction




• Narrow band RF detector• Wake-On-Wireless [Shieh, et

al]– Concept of ‘smart brick’ and

‘mini brick’– Separate frequency channel for

wake-up

• Turducken: Hierarchical power management for mobile devices [Sorber, et al]– Hierarchy of devices– Decomposition of task– Use of WiFi detectors



Summary: Usage scenario for Rural Networking and use of 802.15.4 is unique

Wake-on-Wireless

Turducken Wake-on-LAN

Separate antenna for detection

Yes Yes No

Separate data and trigger channels

Yes No No

Suitable for low cost long distance links

No No Yes

Related Work


Application Scenarios

• VOIP usage in a typical setup– 10-30 calls of 1-2 minutes duration

• Solar Power WiFi (http://www.green-wifi.org/)– Low cost solar powered WiFi grid.

• On-demand data retrieval for bridge monitoring.



Summary: Wake-on-WLAN is useful in energy constrained use of 802.11 equipments


Applications

External Antenna Base node

Node 4Node 3 Node 2 Node 1

IEEE 802.15.4

IEEE 802.11

BriMon Project

With: Hemanth HaridasCSE, IIT Kanpur




Application

BriMon Project





Application

BriMon Project





• Signature pattern based Wake-on-WLAN– Checks for a pattern in a code window– Overcomes the problem of noise triggered false

wake-up– Works in existence of periodic or non periodic

noise.

• Improvements in switching circuit– Latching circuit – Transistor based switch

New Developments• Introduction




• Lack of power save mode in current WiFi Mesh networking hardware

• Novel Wake-on-WLAN mechanism for multi hop remote on-demand wake-up of mesh nodes.

• Substantial power savings using Wake-on-WLAN (> 91%).

• Prototype tested and verified for rural deployment

• Newer applications emerging.

Conclusions• Introduction



Summary: Power management tools are required for current 802.11 based networks deployed in energy constrained scenarios

Wake-on-WLAN Nilesh Mishra, Kameswari Chebrolu, Bhaskaran Raman and Abhinav Pathak IIT Kanpur Wake-on-WLAN Nilesh Mishra, Bhaskaran Raman, Abhinav Pathak.

Documents

abhinav pathak iit kanpur

wlan nilesh mishra

bhaskaran raman

iit kanpur slide

abhinav pathak iit kanpur

abhinav pathak iit kanpur

sensor node

antenna node