Real-Time Deployment of Multihop Relays for Communication Range Extensionfor Communication Range Extension

Nader Moayeri and Michael R. Souryal

Presented at the Precision Personnel Location for First Responders Technology Workshop

Worcester Polytechnic Institute, Worcester, MA

August 4-6 2008August 4 6, 2008

Motivation & ObjectivesMotivation

Public safety operations require reliable rapidly-require reliable, rapidly-deployed communicationsFrequent wireless dead spots encountered inside largeencountered inside large buildings, in underground tunnels, and other difficult radio environmentsradio environments

ObjectivesMaintain network connectivity f li bl i tifor reliable communicationsMinimize impact on user’s mission

ApproachExtend signal coverage through a multihop network that relays data/voice betweenthat relays data/voice between first responders and Incident Command (IC)Relays ( ) are automaticallyRelays ( ) are automatically deployed at appropriate points along path, creating a multihop networkLink layer and routing protocols provide reliable two-way communications, transfer of user vital signs, etc.

Technical ChallengesAutomating the deployment process…When/where to deploy relays?

Incident Command Link

Measurement

Reliable end-to-end transmission

Measurement

Reliable end to end transmission» Link-quality-sensitive route metric» Link and network layer retransmission

Adapting to changes in link quality and topology» Timely route updates

P t l» Power control

Questions & OverviewQuestions to Address

How to measure the quality of a wireless link reliably and efficiently? (in order to determine when to deploy a new relay)(in order to determine when to deploy a new relay)What criteria should be used to trigger deployment?Is real-time on-the-fly deployment feasible? (Will it result in well-connected networks capable of reliable communications?)connected networks capable of reliable communications?)

Remainder of Today’s PresentationApproach to real-time deploymentOverview of the prototypeNext stepse t steps

Live Demo of the Prototype: Tomorrow Morning, Mid-Century

Link Reliability Measurements

Packet Reception Rate vs. Average Received Signal Strength at 900 MHz

Packet Reception Rate vs. Average Signal-to-Noise Ratio at 2.4 GHz

70

80

90

100

%)

70

80

90

100

%)

Theoretical, AWGNExperimental

40

50

60

70

rece

ptio

n ra

te (%

40

50

60

70

rece

ptio

n ra

te (%

10

20

30

40

Pac

ket

10

20

30

40

Pac

ket

-105 -95 -85 -75 -65 -550

Average received signal strength (dBm)0 10 20 30 40 50 60 70 80

0

Average signal-to-noise ratio (dB)

C ll t d fi d t l i l fl f ffi b ildi• Collected over a fixed topology on single floor of an office building• Clear threshold ⇒ RSS/SNR as indicator of link reliability

Mobile Link Measurements

SNR vs. Time on a 2.4 GHz Mobile Link

RSS vs. Time on a 900 MHz Mobile Link

-50

-40

(dB

m)

70

80

90

100Instaneous SNRFiltered SNR

-70

-60

sign

al s

treng

th

40

50

60

70

SN

R (d

B)

-90

-80

Rec

eive

d

10

20

30

40

• Measured by a mobile rx moving at fixed velocity down an office corridor

0 500 1000 1500 2000 2500 3000 3500 4000-100

Packet number10 15 20 25 30 35 400

Time (s)

• Measured by a mobile rx moving at fixed velocity down an office corridor• ⇒ Approaches needed to tolerate multipath fading

Overview of Deployment Algorithm

Mobile Node

Probe ACKsBase Node Channel

Probes

Deployed Relays

1 M bil d b1. Mobile node probes channel every Δ sec.

2. Measures SNR of each

Deployment triggered

Avg.

SN

R

(for 2nd order link diversity)

Probe ACK (bidirectional)

3. If average SNR of each responding relay is less

A

Sth

responding relay is less than a threshold, trigger deployment of new relay. Time

2.4 GHz Prototype BreadcrumbLithium polymer battery &battery & charger

Gumstix motherboardGumstix motherboard400 MHz Linux computer16 MB Flash64 MB SDRAM

Wifistix expansion boardIEEE 802.11b/gO d i64 MB SDRAM

8 cm × 2 cmOpen source driver8 cm × 2 cm

Deployment MonitorNext relay to be deployed continuously probes and measures link quality to its neighbors.

Deployment monitor on mobile display:

When that relay is deployed, next relay is set to probe.

2.4 GHz Prototype System FeaturesApplications

Two-way voice between Base and Mobile NodeC f ’Continuous monitoring of Mobile Node’s sensorsDisplay approximate location of Mobile Node– using RFID-assisted inertial navigationusing RFID assisted inertial navigation

Video

Technical FeaturesTechnical FeaturesRapid link measurementReal-time deployment w/ local placement assistanceIEEE 802.11 PHY/MACOLSR routing w/ ETX metricIP supportIP support

Ongoing and Future Work

Systematic study of link quality measurement techniques

Routing protocol improvementsRouting protocol improvements– Smoother route transitions– Incorporate new link quality metricsIncorporate new link quality metrics

Feasibility of image/video over deployed multihop network

N t k i l ti f iti di l t kNetwork simulation of cognitive radio relay networks

PublicationsM. T. Refaei, M. R. Souryal, and N. Moayeri, "Interference avoidance in rapidly deployed wireless ad hoc incident area networks," Proc. 2nd IEEE Workshop on Mission Critical Networkingnetworks, Proc. 2 IEEE Workshop on Mission Critical Networking (MCN), April 2008.

M. R. Souryal, J. Geissbuehler, L. E. Miller, and N. Moayeri, "Real-time deployment of multihop relays for range extension " Proc ACMtime deployment of multihop relays for range extension, Proc. ACM MobiSys, June 2007.

M. R. Souryal, L. Klein-Berndt, L. E. Miller, and N. Moayeri, "Link assessment in an indoor 802.11 network," Proc. IEEE WCNC, April 2006.

A. Wapf and M. R. Souryal, "Measuring indoor mobile wireless link p y , gquality," draft.

B k SlidBackup Slides onRelay Deployment

Deployment Algorithm Tradeoffs

Δ Probe period 100 msN A i filt l th 20

Deployment Parameters

N Averaging filter length 20

Sth Threshold RSS -80 dBm

Δ N 4 Δ 100

10

15Δ × N = 4 sec

S th

(dB

)

10

15

S th

(dB

)

Δ = 100 ms

0

5

men

t RS

S –

0

5

men

t RS

S –

-10

-5

ost-d

eplo

ym

-10

-5

ost-d

eplo

ym

100, 40 200, 20 500, 8-15

Probe period, filter length (Δ, N)

Po

5 10 20 40-15

( )Filter length (N)

Po

2.4 GHz Symmetric Link Measurements

80• Fixed STA “A”, mobile STA “B”

Bi-Directional Filtered SNR Measurements of a Symmetric Mobile Link

60

70 • “B” carried down 110 m corridor and back at ~ 1.2 m/s

• Sampling period: 100 ms• Filter: Uniform moving average of

50

60

SN

R (d

B) • Filter: Uniform moving average of

last 20 samples (2 s filter)• Data rate: 2 Mbps

30

40

Filte

red

S

20

30

Link A->BLi k B >A

0 20 40 60 80 100 120 140 160 18010

Time (s)

Link B->A

2.4 GHz Asymmetric Link Measurements

80• Fixed STA “A”, mobile STA “B”

Bi-Directional Filtered SNR Measurements of an Asymmetric Mobile Link

60

70 • “B” carried down 110 m corridor and back at ~ 1.2 m/s

• Sampling period: 100 ms• Filter: Uniform moving average of

50

60

SN

R (d

B)

• Filter: Uniform moving average of last 20 samples (2 s filter)

• Data rate: 2 Mbps

30

40

Filte

red

S

20

30

Link A->BLi k B >A

0 20 40 60 80 100 120 140 160 18010

Time (s)

Link B->A

Backup Slides on900 MHz Prototypeyp

900 MHz Experimental PlatformCrossbow MICA2 Mote (MPR400CB)– ChipCon CC1000 transceiver at 916 MHz– 8-bit ATMega128L 7 37 MHz processor8 bit ATMega128L 7.37 MHz processor– 128 kB program memory, 4 kB SRAM– Powered by 2 AA batteries– 5 dBm max RF power5 dBm max. RF power

Multi-Sensor Module (MTS310)Multi-Sensor Module (MTS310)– Light, Temperature – Acoustic, Sounder – 2-Axis Accelerometer– 2-Axis Accelerometer – 2-Axis Magnetometer

900 MHz Prototype SystemApplications

Continuous monitoring of Mobile Node’s sensors

Base Node Mobile Node

Two-way text messaging between Base and MobileDisplay approximate p y pplocation of Mobile Node– using RFID location tags

T h i l F tTechnical FeaturesRapid link measurement w/ adaptive probingReal-time deployment w/ local placement assistanceModified DSDV routing with

Relays

link quality metricPower control

Backup Slides on2.4 GHz Prototype Testingyp g

Test in NIST AMLMarch 24 & 26, 2008Buildings 217, 218, and 219

Deployment Example8

Bldg 219-SB

Bldg 217 Stairwell 3

8 relays deployed

IEEE 802.11 2-Mbps data rate 9

8

G5Stairwell 3

IC

OLSR– HELLO period 0.5 s

D i d l tB

6

During deployment:– 28 kbps full-duplex VoIP

call between IC and FR SB7

– Ping every second

After deployment:– 10 MB file transfer

11 10

– 10 MB file transfer– Audio recording

12 FR12

Bldg 218-SB

FR

During Deployment

9

1011% Packet Loss Rate

7

8

5

6

ber o

f hop

s

3

4

Num

0

1

2

100 200 300 400 500 600 7000

Time (s)

During Deployment: Packet Losses

9

1011% Packet Loss Rate

7

8

5

6

ber o

f hop

s

3

4

Num

0

1

2

100 200 300 400 500 600 7000

Time (s)

During Deployment: Round-Trip Time

100AverageMedian

60

80

ime

(ms)

40

60

an ro

und-

trip

t

20

vera

ge/m

edia

20

0

Av

0 1 2 3 4 5 6 7 8 9 10-20

Number of hops

File Transfer8

Bldg 219-SB

Bldg 217 Stairwell 3

9-hop route

10 MB file 9

8

G5Stairwell 3

IC

Transmission time: 8 min 3 sAverage throughput: 166 kbps

– Peak throughput: 232 kbpsB

6

– Peak throughput: 232 kbps

Simultaneous ping:– Average RTT: 173 ms

SB7

– Packet loss rate: 36% 11 10

12 FR12

Bldg 218-SB

FR

8

Audio Bldg 219-SB

Bldg 217 Stairwell 3

9

8

G5Reading of Gettysburg AddressLinphone VoIP connection with 16 kHz Speex codec at 28 kbps

Stairwell 3IC

B6

16 kHz Speex codec at 28 kbps

Simultaneous Ping Results

orig recv

SB7

9

s 40/58/9411 10

7

8

umbe

r of h

ops

35/54/100

40/58/94

126

7Nu

29/41/55Min/Avg/Max RTTs (ms) 12

Bldg 218-SB180 200 220 240 260 280 300

Time (s)

Min/Avg/Max RTTs (ms)