21
Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-Based WSN Transport Protocol Nestor M. C. Tiglao , António M. Grilo INESC-ID/Instituto Superior Técnico Lisbon, Portugal 6 March 2013 DRCN 2013, Budapest, Hungary

On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Embed Size (px)

DESCRIPTION

Paper presented at DRCN 2013

Citation preview

Page 1: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

On the Optimization and Comparative

Evaluation of a Reliable and Efficient

Caching-Based WSN Transport Protocol

Nestor M. C. Tiglao, António M. Grilo

INESC-ID/Instituto Superior TécnicoLisbon, Portugal

6 March 2013DRCN 2013, Budapest, Hungary

Page 2: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Outline

1 Introduction

2 Related workCaching-based WSN TransportDTSN

3 Proposed MechanismsNACK RepairAdaptive MAC RetryTransmission Window Optimization

4 Performance Evaluation

5 Conclusion

Page 3: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Wireless Sensor Network

Composed of small, resource-constrained wireless devices

Multi-hop operation

Transport protocol: reliability, congestion control,energy-efficiency

Page 4: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Motivation

Develop simple mechanisms that can be implemented inconstrained devices (i.e., O(1) complexity)

Explore novel approaches in the transport layer

Leverage on intermediate caching to improve performance

Page 5: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Caching-based Transport Protocols

Pump Slowly, Fetch Quickly (PSFQ, 2002)

sink-to-sensor, hop-by-hop reliability, designed for codeupdate, uses broadcast

Reliable Multi-Segment Transport (RSMT, 2003)

end-to-end reliability, uses NACKs, timer-driver lossdetection

Distributed TCP Caching (DTC, 2004)

caching TCP segments and retransmitting segmentslocal in case of packet loss

TCP Support for Sensor Networks (TSS, 2007)

not forward a cached TCP segment until the next-hophas received all previous segments (backpressure)

Distributed Transport for Sensor Networks (DTSN, 2007)

Page 6: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

DTSN

Basic DTSN

Enhanced DTSN

Page 7: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Cross-Layer Approach

Page 8: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Enhanced NACK Repair Mechanism

RNACK Procedure

procedure pkt_recv(pkt)...if (!rpending_ && seqno! =next_) then

repseqno_ ← seqno

rpending_ ← 1 ⊲ raise Repair PendingSend RNACK (seqno)

else

do nothingend if

if (rpending_ && seqno==repseqno_) then

rpending_ ← 0 ⊲ clear Repair Pendingnext_ ← maxseen_ + 1 ⊲ update next_

end if

if (seqno > maxseen_) then

maxseen_ ← seqno ⊲ update maxseen_

end if

...end procedure

Example of the Enhanced NACKRepair Mechanism

Page 9: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Adaptive MAC Retry Limit

r ← max

{

3, ⌊log Π− log p

log p⌋

}

R = 1−Π

Π is the Frame Error Rate(FER)p is the physical layer frame error rater is the MAC retry limitR is the desired MAC layer reliability

0 0.2 0.4 0.6 0.8 10

10

20

30

40

50

60

FER

r

Π=0.8

Π=0.9

Π=0.95

MAC retry limit value, r , forvarious MAC reliability levels

R FER≤0.3 FER=0.5 FER=0.7

80% 3 3 490% 3 3 695% 3 4 8

Page 10: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Transmission Window Optimization

Dynamic Window

Additive Increase Multiplicative, Decrease (AIMD)algorithm (cwnd in TCP)inefficient in wireless networks

Fixed Window

based on the bandwidth-delay product, i.e., W =n4

where n = number of hopsHow about caching-based protocols?

Page 11: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Setup

Simulation Parameters

Parameter Value

Network topology Linear chainPacket size 500 bytesNumber of packets(pktno) 500DTSN EAR interval 200 msecRouting protocol StaticMAC protocol 802.11bMAC retry limit (default) 3 (default)PHY error model Binary Symmetric ChannelMax. simulation time 2,000 secondsSimulator ns-2.31

Assumptions:Routing topology is stable

Cross-layer information isavailable

Scenario 1: Global Hotspot

Scenario 2: Localized Hotspot

Page 12: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

DTSN Transmission Window OptimizationGoodput

AWopt = [CS , CS + ∆], ∆ = 10

2 8 10 20 30 40 500

20

40

60

80

100

120

140

Acknowledgment Window (AW) (in packets)

Goo

dput

(in

pac

kets

/sec

)

FER=0 FER=0.1 FER=0.3 FER=0.5 FER=0.7

(a) CS=10

2 8 10 20 30 40 500

20

40

60

80

100

120

140

Acknowledgment Window (AW) (in packets)

Goo

dput

(in

pac

kets

/sec

)

FER=0 FER=0.1 FER=0.3 FER=0.5 FER=0.7

(b) CS=20

Scenario 1 – Goodput, as a function of AW

Page 13: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

DTSN Transmission Window OptimizationTransmission Cost

tx_cost =Ndata + Nack + Nnack + Nmack

pktno

2 8 10 20 30 40 500

50

100

150

200

250

Acknowledgment Window (AW) (in packets)

Tra

nsm

issi

on C

ost

FER=0 FER=0.1 FER=0.3 FER=0.5 FER=0.7

(a) CS=10

2 8 10 20 30 40 500

50

100

150

200

250

Acknowledgment Window (AW) (in packets)

Tra

nsm

issi

on C

ost

FER=0 FER=0.1 FER=0.3 FER=0.5 FER=0.7

(b) CS=20

Scenario 1 – Transmission Cost, as a function of AW

Page 14: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Performance Analysis

Protocols considered:

DTPA – The DTPA protocol, W = BDP(n) + 3

DTPA-CWL – The DTPA protocol, W = BDP(n)

DTSN+ – The DTSN protocol with the proposedenhanced NACK repair and adaptive MAC retry limitmechanisms

TCP− – The TCP protocol without the RTO exponentialbackoff

Page 15: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Performance AnalysisGoodput

Scenario 1

0 0.10 0.30 0.50 0.700

20

40

60

80

100

120

140

Frame Error Rate

Goo

dput

(in

pac

kets

/sec

)

DTPA−BDPDTPATCP−

DTSN+

Performance Gain of DTSN+

FER DTPA-CWL DTPA TCP−

0 96% 129% 88%0.10 51% 138% 59%0.30 71% 75% 137%0.50 720% 723% 1221%0.70 ∞ ∞ ∞

Scenario 2

0 0.10 0.30 0.50 0.700

20

40

60

80

100

120

140

Frame Error Rate

Goo

dput

(in

pac

kets

/sec

)

DTPA−BDPDTPATCP−

DTSN+

Performance Gain of DTSN+

FER DTPA-CWL DTPA TCP−

0 96% 129% 88%0.10 87% 135% 81%0.30 67% 123% 69%0.50 100% 92% 239%0.70 346% 266% 883%

Page 16: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Performance AnalysisTransmission Cost

Scenario 1

0 0.10 0.30 0.50 0.700

20

40

60

80

100

120

140

160

Frame Error Rate

Tra

nsm

issi

on C

ost

DTPA−BDPDTPATCP−

DTSN+

Performance Gain of DTSN+

FER DTPA-CWL DTPA TCP−

0 29% 58% 29%0.10 20% 65% 21%0.30 19% 54% 19%0.50 39% 54% 49%0.70 ∞ ∞ ∞

Scenario 2

0 0.10 0.30 0.50 0.700

20

40

60

80

100

120

140

160

Frame Error Rate

Tra

nsm

issi

on C

ost

DTPA−BDPDTPATCP−

DTSN+

Performance Gain of DTSN+

FER DTPA-CWL DTPA TCP−

0 29% 58% 29%0.10 28% 60% 28%0.30 23% 63% 22%0.50 25% 57% 25%0.70 31% 48% 49%

Page 17: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Performance AnalysisTCP cwnd Evolution

100 110 120 1300

2

4

6

8

10FER=0

Time (in sec)cw

nd (

in p

kt)

100 110 120 1300

2

4

6

8

10FER=0.1

Time (in sec)

cwnd

(in

pkt

)

100 110 120 1300

2

4

6

8

10FER=0.3

Time (in sec)

cwnd

(in

pkt

)

100 110 120 1300

2

4

6

8

10FER=0.5

Time (in sec)

cwnd

(in

pkt

)

100 110 120 1300

2

4

6

8

10FER=0.7

Time (in sec)

cwnd

(in

pkt

)

Scenario 1 – Evolution of TCP cwnd

Page 18: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Performance AnalysisPacket Reception

100 102 104 106 108 1100

50

100

150

200

250

300

350

400

450

500

Time (in seconds)

Seq

uenc

e N

umbe

r

DTPA−CWLDTPATCP−

DTSN+

(a) FER=0

100 105 110 1150

50

100

150

200

250

300

350

400

450

500

Time (in seconds)

Seq

uenc

e N

umbe

r

DTPA−CWLDTPATCP−

DTSN+

(b) FER=0.1

100 105 110 115 120 125 1300

50

100

150

200

250

300

350

400

450

500

Time (in seconds)

Seq

uenc

e N

umbe

r

DTPA−CWLDTPATCP−

DTSN+

(c) FER=0.3

100 150 200 250 300 3500

50

100

150

200

250

300

350

400

450

500

Time (in seconds)

Seq

uenc

e N

umbe

r

DTPA−CWLDTPATCP−

DTSN+

(d) FER=0.5

Scenario 1 – Packet Reception

Page 19: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

Conclusion

Transmission window and loss recovery semantics forcaching-based transport mechanisms need to be optimized

We have proposed the following mechanisms

enhanced NACK recoveryadaptive MAC retry limitoptimal DTSN transmission window

DTSN+ significantly outperforms TCP and DTPA interms of goodput and energy-efficiency

Future work

consider more complex and dynamic network scenariosstudy performance in presence of network congestion

Page 20: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

References

1 IEEE Standard for Information Technology Part 15.4: Wireless Medium Access Control (MAC) andPhysical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (LR-WPANs), IEEEStd. 802.15.4-2006.

2 C. Wang, K. Sohraby, B. Li, M. Daneshmand, and Y. Hu, "A survey of transport protocols for wirelesssensor networks," IEEE Network, vol. 20, no. 3, pp. 34-40, May-June 2006

3 F. Stann and J. Heidemann, ”Rmst: reliable data transport in sensor networks,” in Proceedings of the FirstIEEE International Workshop on Sensor Network Protocols and Applications, May 2003, pp. 102-112.

4 C.-Y. Wan, A. T. Campbell, and L. Krishnamurthy, "Psfq: a reliable transport protocol for wireless sensornetworks," in Proceedings of the 1st ACM international workshop on Wireless sensor networks andapplications, ser. WSNA ’02. New York, NY, USA: ACM, 2002, pp. 1-11.

5 B. Marchi, A. Grilo, and M. Nunes, "Dtsn: Distributed transport for sensor networks," in 12th IEEESymposium on Computers and Commu- nications. ISCC 2007, July 2007, pp. 165-172.

6 O. Akan and I. Akyildiz, "Event-to-sink reliable transport in wireless sensor networks," IEEE/ACMTransactions on Networking, vol. 13, no. 5, pp. 1003-1016, Oct. 2005.

7 X. Li, P.-Y. Kong, and K.-C. Chua, "Dtpa: A reliable datagram transport protocol over ad hoc networks,"IEEE Transactions on Mobile Computing, vol. 7, no. 10, pp. 1285-1294, Oct. 2008.

8 F. Shaikh, A. Khelil, A. Ali, and N. Suri, "Trccit: Tunable reliability with congestion control forinformation transport in wireless sensor networks," in The 5th Annual ICST Wireless Internet Conference(WICON),March 2010, pp. 1-9.

9 A. Dunkels, J. Alonso, T. Voigt, and H. Ritter, "Distributed tcp caching for wireless sensor networks," inProceedings of the 3rd Annual Mediterranean Ad-Hoc Networks Workshop, 2004.

10 K. Chen, Y. Xue, S. H. Shah, and K. Nahrstedt, "Understanding bandwidth-delay product in mobile adhoc networks," Comput. Commun., vol. 27, no. 10, pp. 923-934, Jun. 2004.

11 N. M. C. Tiglao and A. M. Grilo, "An analytical model for transport layer caching in wireless sensornetworks," Performance Evaluation, vol. 69, no. 5, pp. 227-245, 2012.

12 –, "Cross-layer caching based optimization for wireless multimedia sensor networks," in 8th IEEEInternational Conference on Wireless and Mobile Computing, Networking and Communications. WiMob2012. Oct. 2012, pp. 697-704.

13 "The network simulator - ns-2," http://www.isi.edu/nsnam/ns/.

Page 21: On the Optimization and Comparative Evaluation of a Reliable and Efficient Caching-based WSN Transport Protocol

Introduction Related work Proposed Mechanisms Performance Evaluation Conclusion

End

Thank you for your attention!