Prabal Dutta , Stephen Dawson-Haggerty, Yin Chen, Chieh -Jan (Mike) Liang, and Andreas Terzis

Post on 22-Feb-2016

45 views 0 download

Tags:

description

Design and Evaluation of a Versatile and Efficient Receiver-Initiated Link Layer for Low-Power Wireless. Prabal Dutta , Stephen Dawson-Haggerty, Yin Chen, Chieh -Jan (Mike) Liang, and Andreas Terzis Sensys 2010 Presenter: SY. Outline. Introduction A-MAC primitive Implementation - PowerPoint PPT Presentation

Transcript of Prabal Dutta , Stephen Dawson-Haggerty, Yin Chen, Chieh -Jan (Mike) Liang, and Andreas Terzis

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Design and Evaluation of a Versatile and Efficient Receiver-Initiated Link Layer for Low-Power Wireless

Prabal Dutta, Stephen Dawson-Haggerty, Yin Chen,Chieh-Jan (Mike) Liang, and Andreas Terzis

Sensys 2010

Presenter: SY

Network and Systems Laboratorynslab.ee.ntu.edu.tw

OutlineIntroductionA-MAC primitiveImplementationBackcast evaluationHigh level evaluationDrawbacksConclusion

Network and Systems Laboratorynslab.ee.ntu.edu.tw

3

A sender-initiated MAC:Sender triggers communications by transmitting a data

Receiver

Sender Listen D

DListen

Network and Systems Laboratorynslab.ee.ntu.edu.tw

4

Low-power listening (LPL) with a sender-initiated MAC

Preamble DSender

Receiver D

Tlisten Noise

Overhearing/noise adds significant unpredictability to node lifetime

Network and Systems Laboratorynslab.ee.ntu.edu.tw

5

A receiver-initiated MAC:Receiver triggers exchange by transmitting a probe

Receiver

Sender PListen D

P D

Network and Systems Laboratorynslab.ee.ntu.edu.tw

6

Receiver-initiated services -- benefitsHandle hidden terminals better than sender-initiated

ones

Support asynchronous communication w/o long-preambles

Support extremely low duty cycles or high data rates

Support many low-power services Wakeup (“LPP”, Musaloiu-E. et al., IPSN’08) Discovery (“Disco”, Dutta et al., Sensys’08) Unicast (“RI-MAC”, Sun et al., Sensys’08) Broadcast (“ADB”, Sun et al., Sensys’09) Pollcast (“Pollcast”, Demirbas et al., INFOCOM’08) Anycast (“Backcast”, Dutta et al., HotNets’08)

Network and Systems Laboratorynslab.ee.ntu.edu.tw

7

Receiver-initiated services -- drawbacksProbe (LPP) is more expensive than channel sample (LPL)

Baseline power is higher

Frequent probe transmissionsCould congest channel & increase latencyCould disrupt ongoing communicationsChannel usage scales with node density rather than

traffic

Services use incompatible probe semanticsMakes concurrent use of services difficultSupporting multiple, incompatible probes increases

power

Network and Systems Laboratorynslab.ee.ntu.edu.tw

8

The probe incompatibility mess

Probes use hardware acknowledgementsProbes do not use hardware acknowledgementsProbes include only receiver-specific dataProbes include sender-specific data tooProbes include contention windowsProbes do not include contention windows

Pollcast

RI-MACLPP

Backcast

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Is it possible to design a general-purpose,

yet efficient, receiver-initiated link layer?

9

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Most consequential decision a low-power MAC makes:stay awake or go to sleep?

Preamble DTX

RX D

Tlisten Noise

TX

RX

P

P

DATA

DATA

Listen

Sender-Initiated: Channel Sampling

Receiver-Initiated: Channel Probing

P10

Network and Systems Laboratorynslab.ee.ntu.edu.tw

11

Solving the synchronization problem with BackcastA link-layer frame exchange in which:

A single radio PROBE frame transmission Triggers zero or more identical ACK frames Transmitted with tight timing tolerance So there is minimum inter-symbol interference And ACKs collide non-destructively at the receiver

P ATX

RX P A

P ATXYou sh

ould be skeptica

l that th

is idea might

work

P. Dutta, R. Musaloiu-E., I. Stoica, A. Terzis, “Wireless ACK Collisions Not Considered Harmful”,HotNets-VII, October, 2008, Alberta, BC, Canada

Network and Systems Laboratorynslab.ee.ntu.edu.tw

OutlineIntroductionA-MAC primitiveImplementationBackcast evaluationHigh level evaluationDrawbacksConclusion

Network and Systems Laboratorynslab.ee.ntu.edu.tw

13

A-MAC: An 802.15.4 receiver-initiated link layer

P ASender

Receiver P A

DATA

DATA

Max data packet

4.256 ms

ACK transmission time 352 µs

RXTX turnaround time: 192 µs

P

P

Listen

Network and Systems Laboratorynslab.ee.ntu.edu.tw

14

A-MAC’s contention mechanism

Receiver

Sender

Sender P AListen D P-CW

P AListen D P-CW

P A D P-CW D

BO

D

frame collisionBackcast

Network and Systems Laboratorynslab.ee.ntu.edu.tw

15

A-MAC’s parallel multichannel data transfersuse control, data (1), and data (2) channels

P ASender 1

Receiver 1 P A

DATA

DATA

P

P

Listen

P ASender 2

Receiver 2 P A

DATA

DATA

P

P

Listen

Network and Systems Laboratorynslab.ee.ntu.edu.tw

OutlineIntroductionA-MAC primitiveImplementation

UnicastBroadcastPollcastWakeup

Backcast evaluationHigh layer evaluationConclusion

Network and Systems Laboratorynslab.ee.ntu.edu.tw

17

Unicast

P ANode 2(Receiver)

Node 3(Sender)

P ANode 1(Sender) Listen

DDST=0x8002SRC=0x0002

D P

P L

MAC=0x8002

P AListen D P-CWMAC=0x8002

P AListen D P-CWMAC=0x8002

P A D P-CW D

BO

D

DST=0x8002SRC=0x0002ACK=0x0023FRM=0x0001

DST=0x0002SRC=0x0001SEQ=0x23

frame collisionBackcast

Network and Systems Laboratorynslab.ee.ntu.edu.tw

18

Broadcast

TX 1

RX 2

RX 3

RX 4

Listen P A Listen P A Listen

P A

P A

Backcast

auto-ack=onaddr-recog=off

D

D

P

P

D

D

P

P

P A D P

P A D P

DST=0x8002SRC=0x0002

DST=0x8002SRC=0x0002ACK=0x0023FRM=0x0001

DST=0x0002SRC=0x0001SEQ=0x23

offoff

onoff

BackcastBackcast

Network and Systems Laboratorynslab.ee.ntu.edu.tw

19

Wakeup

P ANode 2

Node 3

Node 4

P ANode 1

Node 5

Listen

Listen P A Listen P A Listen

P A Listen

Listen

P A

P A

Listen P A

P

P

A

A

Backcast

DST=0xFFFFSRC=0x0002

Network and Systems Laboratorynslab.ee.ntu.edu.tw

20

Pollcast

Node 2(Receiver)

Node 3(Sender)

Node 1(Sender) PredEvent

PredEvent

Pred

M. Demirbas, O. Soysal, and M. Hussain,“A Single-Hop Collaborative Feedback Primitive for Wireless Sensor Networks”,INFOCOM’08, April , 2008, Phoenix, AZ

ListenMAC=0x8765

ListenMAC=0x8765

Listen

P A

P A

P A

Backcast

DST=0xFFFFSRC=0x0002

PRED=elephantMAC=0x8765

EventDST=0x8765

P+Pred

P+Pred

P+PredDST=0xFFFFSRC=0x0002

PRED=elephant

A

A

A

Network and Systems Laboratorynslab.ee.ntu.edu.tw

OutlineIntroductionA-MAC primitiveImplementationBackcast evaluationHigh level evaluationDrawbacksConclusion

Network and Systems Laboratorynslab.ee.ntu.edu.tw

22

Evaluating Backcast: equal-power, equal-path delaySetup Methodology

Platform: Telos moteProtocol: IEEE 802.15.4Radio: Texas Instruments

CC2420Experiment

8 responder nodes Connect w/ splitter/attenuator Turned on sequentially Transmit 100 packets 125 ms inter-packet interval Log

RSSI: received signal strength LQI: chip correlation rate ARR: ACK reception rate

Network and Systems Laboratorynslab.ee.ntu.edu.tw

23

As the number of colliding ACKsgoes from one to eight…

But, for two nodes, LQI exhibits outliers and a lower median

Median RSSI increaseslogarithmically

Median LQI remains nearly constant but is more left-tailedACK reception rate stays practically constant

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Backcast scales to a large number of nodes

24

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Energy cost of A-MAC primitives:probe, receive, transmit, and listen

25

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Idle listening cost: under heavy interference

26

Network and Systems Laboratorynslab.ee.ntu.edu.tw

OutlineIntroductionA-MAC primitiveImplementationBackcast evaluationHigh level evaluationDrawbacksConclusion

Network and Systems Laboratorynslab.ee.ntu.edu.tw

A-MAC offers modest unicast performance

28

R

S SS S

Collision Domain

Network and Systems Laboratorynslab.ee.ntu.edu.tw

A-MAC supports multiple parallel unicast flows

29

RS

RS

RS

CollisionDomain

Network and Systems Laboratorynslab.ee.ntu.edu.tw

A-MAC wakes up the network fasterand more efficiently than LPL (Flash) flooding

30

Faster Wakeup

Fewer Packets

A-MAC

LPL (Flash)

A-MACLPL (Flash)

Network and Systems Laboratorynslab.ee.ntu.edu.tw

A-MAC wakeup works well at low duty cycles

31“Wakeup Latency” is

normalized to the probe interval

Tprobe = 4,000 ms

Pavg = 63 µWIavg = 21 µA

N = 59

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Supporting the Collection Tree Protocol (CTP),A-MAC beats LPL on nearly every figure of merit

32

R

S

S

SS

SS

S SN = 59Tdata = 60 sTprobe = 500 ms

Network and Systems Laboratorynslab.ee.ntu.edu.tw

The effect of interference on idle listening:Sampling, Probing, and Backcast

33

Sampling

Channel 18 Channel 26

ProbingBackcast

Sampling

Probing

Backcast

Background noise/traffic/noise in an office environment

Network and Systems Laboratorynslab.ee.ntu.edu.tw

OutlineIntroductionA-MAC primitiveImplementationBackcast evaluationHigh level evaluationDrawbacksConclusion

Network and Systems Laboratorynslab.ee.ntu.edu.tw

35

Primitive

P ASender

Receiver P A

DATA

DATA

Max data packet

4.256 ms

ACK transmission time 352 µs

RXTX turnaround time: 192 µs

Network and Systems Laboratorynslab.ee.ntu.edu.tw

Backcast degrades when path delay differences exceeds approximately 500 ns (500 ft free space)

36

Network and Systems Laboratorynslab.ee.ntu.edu.tw

A-MAC single channel PDR degrades with high node density and high probe frequency

37

S

RS

S

CollisionDomain

S S

S S S S

S

SSSS

S S

S S

S

Network and Systems Laboratorynslab.ee.ntu.edu.tw

38

ConclusionBackcast provides a new synchronization primitive

Can be implemented using a DATA/ACK frame exchange Works even with a 8, 12, 94 colliding ACK frames Faster, more efficient, and more robust than LPL, LPP

A-MAC augments Backcast to implement Unicast Broadcast Network wakeup Robust pollcast

Results show Higher packet delivery ratios Lower duty cycles Better throughput (and min/max fairness) Faster network wakeup Higher channel efficiency