+ All Categories
Home > Documents > Routing on Roles - TU Dresden

Routing on Roles - TU Dresden

Date post: 05-Dec-2021
Category:
Upload: others
View: 3 times
Download: 0 times
Share this document with a friend
79
Routing on Roles An Adaptive Approach for DTN Routing José Irigon de Irigon November 30, 2018 1
Transcript
Page 1: Routing on Roles - TU Dresden

Routing on RolesAn Adaptive Approach for DTN Routing

José Irigon de IrigonNovember 30, 2018

1

Page 2: Routing on Roles - TU Dresden

Table of contents

1. DTN Introduction

2. Challenges in DTN routing

3. The need for adaptation

4. Adaptation by means of Roles

2

Page 3: Routing on Roles - TU Dresden

DTN Introduction

Page 4: Routing on Roles - TU Dresden

End-to-end connectivity

Figure 1: End-to-end Connectivity Figure 2: No end-to-end Connectivity

3

Page 5: Routing on Roles - TU Dresden

A communication between static stations

. .

.

.

.

.

SBSA

M1

4

Wireless range do not allow end-to-end connectivity• Contact opportunity

• Knowledge about the topology

• Nodes, Stations, Mobile

Page 6: Routing on Roles - TU Dresden

A communication between static stations

. .

.

.

.

.

SBSA

M1

4

Wireless range do not allow end-to-end connectivity

• Contact opportunity

• Knowledge about the topology

• Nodes, Stations, Mobile

Page 7: Routing on Roles - TU Dresden

A communication between static stations

DST:SB

. .

.

.

.

.

SBSA

M1

4

Wireless range do not allow end-to-end connectivity• Contact opportunity

• Knowledge about the topology

• Nodes, Stations, Mobile

Page 8: Routing on Roles - TU Dresden

A communication between static stations

DST:SB

. .

.

.

.

.

SBSA

M1

4

Wireless range do not allow end-to-end connectivity

• Contact opportunity

• Knowledge about the topology

• Nodes, Stations, Mobile

Page 9: Routing on Roles - TU Dresden

A communication between static stations

DST:SB

DST:SB

. .

.

.

.

.

SBSA

M1

4

Wireless range do not allow end-to-end connectivity• Contact opportunity

• Knowledge about the topology

• Nodes, Stations, Mobile

Page 10: Routing on Roles - TU Dresden

A communication between static stations

DST:SB

DST:SB

. .

.

.

.

.

SBSA

M1

4

Wireless range do not allow end-to-end connectivity• Contact opportunity

• Knowledge about the topology

• Nodes, Stations, Mobile

Page 11: Routing on Roles - TU Dresden

A communication between static stations

DST:SBDST:SB

DST:SB

. .

.

.

.

.

SBSA

M1

4

Wireless range do not allow end-to-end connectivity• Contact opportunity

• Knowledge about the topology

• Nodes, Stations, Mobile

Page 12: Routing on Roles - TU Dresden

Challenge networks

Pictures [1, 2, 3, 4]

5

Ring Road[5]: latency insensitive services with LEO satellitesThe village: DakNet[5], KioskNet[6], TrainNet[7], the Sámis[8, 9, 10, 11, 12]Wild life[13] and Smart Farming [14]The Mars rover communication network

Page 13: Routing on Roles - TU Dresden

Challenge networks

Pictures [1, 2, 3, 4]

5

Ring Road[5]: latency insensitive services with LEO satellites

The village: DakNet[5], KioskNet[6], TrainNet[7], the Sámis[8, 9, 10, 11, 12]Wild life[13] and Smart Farming [14]The Mars rover communication network

Page 14: Routing on Roles - TU Dresden

Challenge networks

Pictures [1, 2, 3, 4]

5

Ring Road[5]: latency insensitive services with LEO satellites

The village: DakNet[5], KioskNet[6], TrainNet[7], the Sámis[8, 9, 10, 11, 12]

Wild life[13] and Smart Farming [14]The Mars rover communication network

Page 15: Routing on Roles - TU Dresden

Challenge networks

Pictures [1, 2, 3, 4]

5

Ring Road[5]: latency insensitive services with LEO satellitesThe village: DakNet[5], KioskNet[6], TrainNet[7], the Sámis[8, 9, 10, 11, 12]

Wild life[13] and Smart Farming [14]

The Mars rover communication network

Page 16: Routing on Roles - TU Dresden

Challenge networks

Pictures [1, 2, 3, 4]

5

Ring Road[5]: latency insensitive services with LEO satellitesThe village: DakNet[5], KioskNet[6], TrainNet[7], the Sámis[8, 9, 10, 11, 12]Wild life[13] and Smart Farming [14]

The Mars rover communication network

Page 17: Routing on Roles - TU Dresden

Challenges in DTN routing

Page 18: Routing on Roles - TU Dresden

A concrete use-case

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

M1

6

• Rovers move over defined paths

Page 19: Routing on Roles - TU Dresden

A concrete use-case

6

Page 20: Routing on Roles - TU Dresden

A concrete use-case

6

Page 21: Routing on Roles - TU Dresden

A concrete use-case

6

Bob wants to send a picture to Alice

Page 22: Routing on Roles - TU Dresden

A concrete use-case

6

Page 23: Routing on Roles - TU Dresden

A concrete use-case

6

Routing in DTNMessage Forwarding:

• Whom to forward?• Single or multiple copies

What are the priorities?• energy consumption?• assure delivery?

Knowledge about the network• How much?• How accurate?

Page 24: Routing on Roles - TU Dresden

The routing decision

ObjectiveRouting

AlgorithmContext

.

.

7

Based on the available topological information, a routing protocolis chosen that uses the network objectives as priorities to decideupon conflicting decisions.

What is known about the topology?

Page 25: Routing on Roles - TU Dresden

The routing decision

Contact

Capability

Historical

InformationPrediction

Contact

Plan (CP)

Latency

Energy

efficiency

Number

of hops

Queueing

delay. . .

Delivery

probability

Replication

Based

Probability

Based

CP Based

ObjectiveRouting

AlgorithmContext

.

.

7

Based on the available topological information, a routing protocolis chosen that uses the network objectives as priorities to decideupon conflicting decisions.

What is known about the topology?

Page 26: Routing on Roles - TU Dresden

The routing decision

Contact

Capability

Historical

InformationPrediction

Contact

Plan (CP)

Latency

Energy

efficiency

Number

of hops

Queueing

delay. . .

Delivery

probability

Replication

Based

Probability

Based

CP Based

ObjectiveRouting

AlgorithmContext

.

.

8

What are the rover priorities?

Page 27: Routing on Roles - TU Dresden

The routing decision

Contact

Capability

Historical

InformationPrediction

Contact

Plan (CP)

Latency

Energy

efficiency

Number

of hops

Queueing

delay. . .

Delivery

probability

Replication

Based

Probability

Based

CP Based

ObjectiveRouting

AlgorithmContext

.

.

9

To which peers should the bundle be forwarded?

Page 28: Routing on Roles - TU Dresden

The routing decision - Replication Based

Contact

Capability

Historical

InformationPrediction

Contact

Plan (CP)

Latency

Energy

efficiency

Number

of hops

Queueing

delay. . .

Delivery

probability

Replication

Based

Probability

Based

CP Based

ObjectiveRouting

AlgorithmContext

.

.

10

Page 29: Routing on Roles - TU Dresden

The routing decision - Replication Based

11

• Bundle is replicated at every contact

• Every node is using the same algorithm

Page 30: Routing on Roles - TU Dresden

The routing decision - Replication Based

11

• Bundle is replicated at every contact

• Every node is using the same algorithm

Page 31: Routing on Roles - TU Dresden

The routing decision - Replication Based

11

Page 32: Routing on Roles - TU Dresden

The routing decision - Replication Based

11

SummaryPositive

• Context independentNegative

• Waste resources• Congestion prone• Poor performance under loador network density

Page 33: Routing on Roles - TU Dresden

The routing decision - Probability Based

Contact

Capability

Historical

InformationPrediction

Contact

Plan (CP)

Latency

Energy

efficiency

Number

of hops

Queueing

delay. . .

Delivery

probability

Replication

Based

Probability

Based

CP Based

ObjectiveRouting

AlgorithmContext

.

.

12

Page 34: Routing on Roles - TU Dresden

The routing decision - Probability Based

elivery predictability is the transitive likelihood of encounter13

• Bundle is replicated to peers that arelikely to encounter the destination

Page 35: Routing on Roles - TU Dresden

The routing decision - Probability Based

elivery predictability is the transitive likelihood of encounter13

ProPHETv2Routing decision based on:

• Peer delivery predictability• Transitive delivery predictability• Aging

Page 36: Routing on Roles - TU Dresden

The routing decision - Probability Based

elivery predictability is the transitive likelihood of encounter13

ProPHETv2Routing decision based on:

• Peer delivery predictability• Transitive delivery predictability• Aging

Page 37: Routing on Roles - TU Dresden

The routing decision - Probability Based

elivery predictability is the transitive likelihood of encounter13

Page 38: Routing on Roles - TU Dresden

The routing decision - Probability Based

elivery predictability is the transitive likelihood of encounter13

SummaryPositive

• Smaller amount of replicas

• Better performance under loadNegative

• Requires a well defined pattern

Page 39: Routing on Roles - TU Dresden

The routing decision - Deterministic

Contact

Capability

Historical

InformationPrediction

Contact

Plan (CP)

Latency

Energy

efficiency

Number

of hops

Queueing

delay. . .

Delivery

probability

Replication

Based

Probability

Based

CP Based

ObjectiveRouting

AlgorithmContext

.

.

14

Page 40: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

Page 41: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

Contact Plan from all devices available

Page 42: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

Contact Plan from all devices availableBuild a multigraph*

Page 43: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

Contact Plan from all devices availableBuild a multigraph*Calculate shortest path

Page 44: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

Page 45: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

The process repeats at every device for each bundle

Page 46: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

Page 47: Routing on Roles - TU Dresden

The routing decision - Deterministic

15

SummaryPositive

• Smallest amount of replicas

• Best performance under loadNegative

• Requires a contact plan

Page 48: Routing on Roles - TU Dresden

A word about congestion in DTN

ObjectiveRouting

AlgorithmContext

.

.

16

• Definition• Why do we care?• What can be done?

”... congestion occurs when resource demands fromusers/applications exceed the network’s available capacity.” [15]

”Generally, it (congestion) occurs when the nodes in the networkbecome overloaded.” [16]

Page 49: Routing on Roles - TU Dresden

A word about congestion in DTN

ObjectiveRouting

AlgorithmContext

.

.

16

”... congestion occurs when resource demands fromusers/applications exceed the network’s available capacity.” [15]

”Generally, it (congestion) occurs when the nodes in the networkbecome overloaded.” [16]

Page 50: Routing on Roles - TU Dresden

Congestion Mitigation

Congestion

ControlCongestion

ControlSelfishness

FairnessQueueing

Dely/Jitter

ObjectiveRouting

AlgorithmContext

.

.

17

Page 51: Routing on Roles - TU Dresden

The need for adaptation

Page 52: Routing on Roles - TU Dresden

Routing adaptation

• At boot up, historical information is not available

• Probabilistic approach provide better results on predictable networks

• A contact plan may be available later on

18

Page 53: Routing on Roles - TU Dresden

Routing adaptation

• At boot up, historical information is not available

• Probabilistic approach provide better results on predictable networks

• A contact plan may be available later on

18

Page 54: Routing on Roles - TU Dresden

Routing adaptation

• At boot up, historical information is not available

• Probabilistic approach provide better results on predictable networks

• A contact plan may be available later on

18

Page 55: Routing on Roles - TU Dresden

Routing adaptation

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

.

.

.

.

M1

19

Page 56: Routing on Roles - TU Dresden

Routing adaptation

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

.

.

.

.

M1

19

An accurate contact plan

...is no assurance against topology changes

Page 57: Routing on Roles - TU Dresden

Routing adaptation

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

.

.

.

.

M1

19

An accurate contact plan...is no assurance against topology changes

Page 58: Routing on Roles - TU Dresden

Routing adaptation

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

.

.

.

.

M1

19

• An alternative path is found

• Contact plan becomes invalid

• Historical information is currently useless

• Switch to a replication-based variant

Page 59: Routing on Roles - TU Dresden

Routing adaptation

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

.

.

.

.

M1

19

• An alternative path is found

• Contact plan becomes invalid

• Historical information is currently useless

• Switch to a replication-based variant

Page 60: Routing on Roles - TU Dresden

Routing adaptation

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

.

.

.

.

M1

19

• An alternative path is found

• Contact plan becomes invalid

• Historical information is currently useless

• Switch to a replication-based variant

Page 61: Routing on Roles - TU Dresden

Routing adaptation

SA SB SC SD SE

SH

SI

SK

SJ

SG

SF

SL

.

.

.

.

M1

19

• The way become free

• Switch to replication-based followed byprobabilistic or schedule-based approach

Page 62: Routing on Roles - TU Dresden

Adaptation by means of Roles

Page 63: Routing on Roles - TU Dresden

Why Roles for this use-case?

• DTN routing algorithms as roles may be implemented andtested in isolation as standalone building blocks

• Constraints can be applied between roles

• Algorithms can be added/extended incrementally

• Test and comparison of new algorithms made easier

20

Page 64: Routing on Roles - TU Dresden

Routing and roles

21

Page 65: Routing on Roles - TU Dresden

Routing and roles

21

Page 66: Routing on Roles - TU Dresden

Routing and roles

21

Page 67: Routing on Roles - TU Dresden

Routing and roles

21

Page 68: Routing on Roles - TU Dresden

Routing and roles

21

Page 69: Routing on Roles - TU Dresden

Routing and roles

21

Page 70: Routing on Roles - TU Dresden

Routing and roles

21

Page 71: Routing on Roles - TU Dresden

Routing and roles

21

Page 72: Routing on Roles - TU Dresden

Routing and roles

21

Page 73: Routing on Roles - TU Dresden

Summary

• Adaptability in DTN is needed at least on start up (learning phase) andunder topology changes

• Choosing the right routing algorithm and respective congestionmitigation mechanism in run time is critical to maximize networkutilization, and should be done based in the current network context.

• The concept of roles offers key features to be used in an adaptiverouting framework:

• possibility to add behavior to unrelated objects• constraints between roles and role groups• run time adaptation

22

Page 74: Routing on Roles - TU Dresden

Thank you

Questions?

23

Page 75: Routing on Roles - TU Dresden

References i

Montry, “Smart Farm.” https://www.wespeakiot.com/robust-sensors-and-the-power-of-the-cloud-the-perfect-recipe-for-smart-farming/,2017.A. Pentland, R. Fletcher, and A. Hasson, “DakNet: RethinkingConnectivity in Developing Nations,” Computer, vol. 37, no. 1, 2004.

B. Rabtsevich and Shutterstock, “Leo Sats.”https://spacewatchme.com/2017/02/geo-leo-hybrid-multigrid-network/.M. D. L. NASA/JPL/Cornell University, “Mars Rover.” https://en.wikipedia.org/wiki/Mars_Exploration_Rover,2003.

24

Page 76: Routing on Roles - TU Dresden

References ii

S. C. Burleigh and E. J. Birrane, “Toward a communicationssatellite network for humanitarian relief,” Proceedings of the 1stInternational Conference on Wireless Technologies forHumanitarian Relief - ACWR ’11, p. 219, 2011.

S. H. Watson, A. A. Thobhani, O. G. Drive, and B. B. Chan, “Designand Implementation of the,” pp. 1633–1644, 2001.

M. Zarafshan-Araki and K. W. Chin, “TrainNet: A transport systemfor delivering non real-time data,” Computer Communications,vol. 33, no. 15, pp. 1850–1863, 2010.

A. Lindgren and A. Doria, “Experiences from deploying a real-lifeDTN system,” 2007 4th Annual IEEE Consumer Communicationsand Networking Conference, CCNC 2007, pp. 217–221, 2007.

25

Page 77: Routing on Roles - TU Dresden

References iii

S. Grasic, Development and Deployment of Delay TolerantNetworks: An Arctic Village Case.2014.K. Heimerl and E. Brewer, “The village base station,” Proceedingsof the 4th ACM Workshop on Networked Systems for DevelopingRegions - NSDR ’10, pp. 1–2, 2010.

A. Lindgren, A. Doria, J. Lindblom, and M. Ek, “Networking in theland of northern lights,” Proceedings of the 2008 ACM workshopon Wireless networks and systems for developing regions -WiNS-DR ’08, pp. 1–7, 2008.

S. Grasic and A. Lindgren, “Revisiting a remote village scenarioand its DTN routing objective,” Computer Communications,vol. 48, pp. 133–140, 2014.

26

Page 78: Routing on Roles - TU Dresden

References iv

P. Juang, H. Oki, Y. Wang, M. Martonosi, P. Peh Li-Shiuan, andD. Rubenstein, “Energy-Efficient Computing for Wildlife Tracking:Design Tradeoffs and Early Experiences with ZebraNet,”Proceedings of the 10th International Conference onArchitectural Support for Programming Languages andOperating Systems (ASPLOS 2002), pp. 96–107, 2002.

C. Kulatunga, L. Shalloo, W. Donnelly, E. Robson, and S. Ivanov,“Opportunistic Wireless Networking for Smart Dairy Farming,” ITProfessional, vol. 19, no. 2, pp. 16–23, 2017.

A. P. Silva, S. Burleigh, C. M. Hirata, and K. Obraczka, “A survey oncongestion control for delay and disruption tolerant networks,”Ad Hoc Networks, vol. 25, no. PB, pp. 480–494, 2015.

27

Page 79: Routing on Roles - TU Dresden

References v

A. Roy, T. Acharya, and S. DasBit, “Quality of service in delaytolerant networks: A survey,” Computer Networks, vol. 130,pp. 121–133, 2018.

28


Recommended