![Page 1: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/1.jpg)
UniversityUniversity of Virginia
1
Flash Flooding: Exploiting the Capture Effect Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networksfor Rapid Flooding in Wireless Sensor Networks
Infocom’09Rio de Janeiro, Brazil
Jiakang Lu and Kamin Whitehouse
Department of Computer Science
University of Virginia
![Page 2: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/2.jpg)
UniversityUniversity of Virginia
2
Classic WSN AlgorithmsClassic WSN Algorithms
• Network floods are common and important operations at the heart of most wireless sensor network algorithms.– Routing tree creation– Time synchronization – Code and data dissemination– Node localization– Group formation
• However, network floods are costly in latency due to …
![Page 3: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/3.jpg)
UniversityUniversity of Virginia
3
A
B
C
D
……neighborhood contentionneighborhood contention
A
B
D
C
E
G
I
H
F
CCA+MAC Delay
Tx
Rx Tx
Rx Tx
Rx TxA
B
D
C
![Page 4: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/4.jpg)
UniversityUniversity of Virginia
4
…… and low-duty cycleand low-duty cycle
A
B
D
C
E
G
I
H
F A
B
C
DRx Tx Tx
Rx
Rx
Tx Tx Tx Tx Tx
Minimal Interpacket Spacing
…
…
…
Wake Up
Sleep
![Page 5: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/5.jpg)
UniversityUniversity of Virginia
5
Related WorkRelated Work• Low-duty cycle CSMA networks
– High latency of an LPL flood• [Polastre 2004], [Buettner 2006]
• Wireless senor networks flooding– Do not explicitly optimize for latency
• [Heinzelman 1999], [Levis 2002], [Hui 2004]
• Real-time communication protocols– Point-to-Point, multicast or data collection
• [He 2003], [Watteyne 2006]
• Rapid wakeup scheduling– Requires phase synchronization
• [Lu 2004], [Li 2005], [Lu 2005], [Keshavarzian 2006]
![Page 6: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/6.jpg)
UniversityUniversity of Virginia
6
Flash OverviewFlash Overview• The Flash flooding protocol exploits the
capture effect to reduce flooding latency by eliminating neighborhood contention– Capture: a radio successfully demodulates one of
multiple overlapping transmissions of the same frequency
– Allow nodes to propagate the message concurrently in a flooding scenario
– Propose three flooding-specific mechanisms to manage transmission concurrency
![Page 7: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/7.jpg)
UniversityUniversity of Virginia
7
OutlineOutline
• Experiment Methodology
• Design of Flash
• Performance evaluation
• Conclusions
![Page 8: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/8.jpg)
UniversityUniversity of Virginia
8
Evaluation MethodologyEvaluation Methodology
• VineLab testbed– 48 Tmote-skys– Office environment
• Trace-based Simulation– Capture-aware simulation framework– Multiple Scales and densities– Statistically verified with the testbed results
![Page 9: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/9.jpg)
UniversityUniversity of Virginia
9
Flash-I: Complete ConcurrencyFlash-I: Complete Concurrency• Carrier sense is completely removed
before transmission– No neighborhood contention
• Tradeoff– Significantly reduce the flooding latency– High network coverage is not guaranteed
TxTx Tx Tx Tx Tx Tx
X-MAC packet trainFlash-I packet train
![Page 10: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/10.jpg)
UniversityUniversity of Virginia
10
Flash-I flooding exampleFlash-I flooding example
A
B
D
C
E
G
I
H
F A
B
C
D
Tx Tx Tx Tx Tx
Rx
Rx
Tx Tx Tx Tx
Tx Tx Tx …
A
B
D
C
Minimal Interpacket Spacing
![Page 11: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/11.jpg)
UniversityUniversity of Virginia
11
Flash-II: Maintained ConcurrencyFlash-II: Maintained Concurrency
• Flash-II achieves low flooding latency while improving the coverage of Flash-I
• Each node has two phases of flooding:1) Flash-I flood
• With no CCA or MAC delay
2) Neighborhood rebroadcast • With CCA and MAC delay (X-MAC flood)• Reach any nodes that missed the first wave
![Page 12: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/12.jpg)
UniversityUniversity of Virginia
12
Flash-II flooding exampleFlash-II flooding example
A
B
D
C
E
G
I
H
F A
B
C
D
Tx Tx Tx Tx Tx
Rx
Rx
Tx Tx Tx Tx
Tx Tx Tx …
A
B
D
C
Phase #1 = Flash-I flood
…
…
CCA and MAC delay before local rebroadcast
![Page 13: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/13.jpg)
UniversityUniversity of Virginia
13
Flash-II flooding exampleFlash-II flooding example
A
B
D
C
E
G
I
H
F A
B
C
DRx
Tx Tx Tx Tx
Tx
Tx
…
A
B
D
C
Phase #2 = Local rebroadcast w/ CCA and MAC dealy
…
…
Tx
Tx Tx Tx
…
…
…
![Page 14: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/14.jpg)
UniversityUniversity of Virginia
14
Flash-II Scale SimulationFlash-II Scale Simulation
75%
![Page 15: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/15.jpg)
UniversityUniversity of Virginia
15
Flash-II Density SimulationFlash-II Density Simulation
70%
![Page 16: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/16.jpg)
UniversityUniversity of Virginia
16
1) a small interpacket spacing (IPS)2) a small CCA before the packet train
Flash-III packet train
Flash-III: Controlled ConcurrencyFlash-III: Controlled Concurrency• A fine balance must be achieved to
exploit the capture in a flood
• Flash-III applies a new technique to sense the amount of transmission concurrency
TxTxTx Tx Tx TxTx Tx Tx TxTxTxTx Tx
IPS
CCA
X-MAC packet train
![Page 17: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/17.jpg)
UniversityUniversity of Virginia
17
A
B
D
C
A
B
C
D
Flash-III flooding exampleFlash-III flooding example
A
B
D
C
E
G
I
H
F
Rx
…
Tx Tx Tx Tx Tx
Tx Tx Tx Tx Tx
Rx Tx Tx Tx Tx
Rx
…
![Page 18: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/18.jpg)
UniversityUniversity of Virginia
18
Flash-III Scale SimulationFlash-III Scale Simulation
75%
![Page 19: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/19.jpg)
UniversityUniversity of Virginia
19
Flash-III Density SimulationFlash-III Density Simulation
80%
![Page 20: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/20.jpg)
UniversityUniversity of Virginia
20
ConclusionsConclusions• Flash is the first network flooding protocol for
wireless networks that explicitly exploits the capture effect to optimize for latency.
• The simplicity of Flash can bring substantial performance improvement in the existing systems and have an immediate and practical impact.
• The empirical study of network-wide capture dynamics and the novel capture-aware simulation framework will inspire new studies on capture in the future.
![Page 21: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/21.jpg)
UniversityUniversity of Virginia
21
Thank youThank you
![Page 22: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/22.jpg)
UniversityUniversity of Virginia
22
Backup slidesBackup slides
![Page 23: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/23.jpg)
UniversityUniversity of Virginia
23
Got D!
cases where capture helpscases where capture helps
A
B
D
C
E
F
G
H
I Got B!
Got D!
![Page 24: University University of Virginia 1 Flash Flooding: Exploiting the Capture Effect for Rapid Flooding in Wireless Sensor Networks Infocom ’ 09 Rio de Janeiro,](https://reader035.vdocuments.us/reader035/viewer/2022070323/56649d9f5503460f94a8a310/html5/thumbnails/24.jpg)
UniversityUniversity of Virginia
24
Got A!
cases where collision happenscases where collision happens
A
B
D
C
E
F
G
H
I
Got B!???
Got D!???