Post on 25-May-2020
transcript
RPL cooperation experimentsusing FIT IoT-LAB testbed
Brandon Foubertbrandon.foubert@inria.fr
Julien Montavontmontavont@unistra.fr
Inria Lille - Nord Europe
ICube - UMR 7357
Journees non-thematiques du GDR Reseaux et Systemes Distribues
January 23, 2020
Outline
1 Scientific contextInternet of ThingsRPL: routing in the IoTInherent issues in RPL
2 Contribution
3 Experimentation
4 Conclusion
Internet of Things (IoT)
Set of constrained objects interconnected with the Internetvia wireless communications
ConstraintsComputation powerMemory storageBattery → limited energy
New usages, new standardsClassic IP protocols not efficient with IoT devicesSpecialized standards from the IEEE and the IETF
Brandon Foubert Sharing is caring 1 / 17
RPL: routing in the IoT [WTB12]
Proactive intra-domaindistance-vector routing protocolDestination Oriented DirectedAcyclic Graph (DODAG)Metrics: Hop count, ExpectedTransmission Count (ETX)...Traffic patterns: multi-point topoint, point to multi-point,point to point
RPL root
Network link
DODAG link
Figure 1: Physical and logical topology
[WTB12] T. Winter et al. RPL: IPv6 Routing Protocol for Low-Power and Lossy Networks. RFC 6550. Mar. 2012
Brandon Foubert Sharing is caring 2 / 17
RPL inherent issues
Border router
X
Network link
Internet
Collect station
DODAG link
Figure 2: Border router failure
Root
DODAG link
Figure 3: Funneling effect [WEC05]
Solution = border router redundancyOrphan nodes redirect traffic to another border routerMultiple exit points → traffic shared between multiple paths
[WEC05] Chieh-Yih Wan et al. “Siphon: Overload Traffic Management Using Multi-radio Virtual Sinks in Sensor Networks”. In:Proceedings of the 3rd International Conference on Embedded Networked Sensor Systems. ACM, 2005
Brandon Foubert Sharing is caring 3 / 17
Related work
BR2BR1
DODAG link
Internet
Network linkInternet link
Central coordinator
Figure 4: Central coordination [NMM16]
Rank 4
RootRank 0
DODAG link
Rank 2
Figure 5: Local load balancing [KKP17]
[NMM16] Quang-Duy Nguyen et al. “RPL Border Router Redundancy in the Internet of Things”. In: Ad-hoc, Mobile, andWireless Networks. Ed. by Nathalie Mitton, Valeria Loscri, and Alexandre Mouradian. Springer International Publishing, 2016.isbn: 978-3-319-40509-4[KKP17] H. S. Kim et al. “Load Balancing Under Heavy Traffic in RPL Routing Protocol for Low Power and Lossy Networks”.In: IEEE Transactions on Mobile Computing 16.4 (Apr. 2017), pp. 964–979. issn: 1536-1233
Brandon Foubert Sharing is caring 4 / 17
Related work
BR2BR1
DODAG link
Internet
Network linkInternet link
Central coordinator
X
Figure 4: Single point of failure [NMM16]
Rank 4
RootRank 0
DODAG link
Rank 2
Figure 5: Local load balancing [KKP17]
[NMM16] Quang-Duy Nguyen et al. “RPL Border Router Redundancy in the Internet of Things”. In: Ad-hoc, Mobile, andWireless Networks. Ed. by Nathalie Mitton, Valeria Loscri, and Alexandre Mouradian. Springer International Publishing, 2016.isbn: 978-3-319-40509-4
[KKP17] H. S. Kim et al. “Load Balancing Under Heavy Traffic in RPL Routing Protocol for Low Power and Lossy Networks”.In: IEEE Transactions on Mobile Computing 16.4 (Apr. 2017), pp. 964–979. issn: 1536-1233
Brandon Foubert Sharing is caring 4 / 17
Related work
BR2BR1
DODAG link
Internet
Network linkInternet link
Central coordinator
X
Figure 4: Single point of failure [NMM16]
Rank 3
RootRank 0
DODAG link
Rank 3
Figure 5: Local load balancing [KKP17]
[NMM16] Quang-Duy Nguyen et al. “RPL Border Router Redundancy in the Internet of Things”. In: Ad-hoc, Mobile, andWireless Networks. Ed. by Nathalie Mitton, Valeria Loscri, and Alexandre Mouradian. Springer International Publishing, 2016.isbn: 978-3-319-40509-4[KKP17] H. S. Kim et al. “Load Balancing Under Heavy Traffic in RPL Routing Protocol for Low Power and Lossy Networks”.In: IEEE Transactions on Mobile Computing 16.4 (Apr. 2017), pp. 964–979. issn: 1536-1233
Brandon Foubert Sharing is caring 4 / 17
Related work
BR2BR1
DODAG link
Internet
Network linkInternet link
Central coordinator
X
Figure 4: Single point of failure [NMM16]
Rank 1
RootRank 0
DODAG link
Rank 6
Figure 5: Network instability [KKP17]
[NMM16] Quang-Duy Nguyen et al. “RPL Border Router Redundancy in the Internet of Things”. In: Ad-hoc, Mobile, andWireless Networks. Ed. by Nathalie Mitton, Valeria Loscri, and Alexandre Mouradian. Springer International Publishing, 2016.isbn: 978-3-319-40509-4
[KKP17] H. S. Kim et al. “Load Balancing Under Heavy Traffic in RPL Routing Protocol for Low Power and Lossy Networks”.In: IEEE Transactions on Mobile Computing 16.4 (Apr. 2017), pp. 964–979. issn: 1536-1233
Brandon Foubert Sharing is caring 4 / 17
Outline
1 Scientific context
2 ContributionConsidered scenarioMultiple border routersLoad balancingMultiple IPv6 prefixes
3 Experimentation
4 Conclusion
Considered scenario
Smart cities: smart street lights,smart health, smart parking, etc.→ colocated networksDifferent Internet serviceprovidersDifferent IPv6 prefixesSame IoT stack
Smart City
Figure 6: Smart cities (from [IEE18])
[IEE18] IEEE smart cities. url: https://beyondstandards.ieee.org/smart-cities/smart-smart-cities/ (visited on08/20/2018)
Brandon Foubert Sharing is caring 5 / 17
Multiple border routers
Redundancy → failure resilience & load sharing between exit points⇒ RPL + distributed virtual DODAG root⇒ Initialization using discovering (e.g. [KLR16])
BR2BR1
BR1 DODAGBR2 DODAG
Internet
Network link
Virtual link
Internet link
Control message
Figure 7: Border router discovering and inter-connexion
[KLR16] M. M. Khan et al. “A multi-sink coordination framework for low power and lossy networks”. In: 2016 InternationalConference on Industrial Informatics and Computer Systems (CIICS). Mar. 2016, pp. 1–5
Brandon Foubert Sharing is caring 6 / 17
Multiple border routers
Redundancy → failure resilience & load sharing between exit points⇒ RPL + distributed virtual DODAG root⇒ Initialization using discovering (e.g. [KLR16])
Overhearing
BR2BR1
BR1 DODAGBR2 DODAG
Internet
Network link
Virtual link
Internet link
Control message
Figure 7: Border router discovering and inter-connexion
[KLR16] M. M. Khan et al. “A multi-sink coordination framework for low power and lossy networks”. In: 2016 InternationalConference on Industrial Informatics and Computer Systems (CIICS). Mar. 2016, pp. 1–5
Brandon Foubert Sharing is caring 6 / 17
Multiple border routers
Redundancy → failure resilience & load sharing between exit points⇒ RPL + distributed virtual DODAG root⇒ Initialization using discovering (e.g. [KLR16])
BR2BR1
BR1 DODAGBR2 DODAG
Internet
Network link
Virtual link
Internet link
Control message
VR
Figure 7: Border router discovering and inter-connexion
[KLR16] M. M. Khan et al. “A multi-sink coordination framework for low power and lossy networks”. In: 2016 InternationalConference on Industrial Informatics and Computer Systems (CIICS). Mar. 2016, pp. 1–5
Brandon Foubert Sharing is caring 6 / 17
Load balancing
Border router redundancy → static (i.e. non-adaptative) load balancing⇒ RPL + explicit redirection:
Multiple RPL instances → border router differentiationColocated networks → nodes set ”redirectable” flagCongested border router → DODAG Redirection Solicitation (DRS)
6
Network link
Internet
54
BR2
BR1
3
12
Control messagesBR1 instance BR2 instance
Figure 8: Redirection of node 4 from BR1 to BR2
Brandon Foubert Sharing is caring 7 / 17
Load balancing
Border router redundancy → static (i.e. non-adaptative) load balancing⇒ RPL + explicit redirection:
Multiple RPL instances → border router differentiationColocated networks → nodes set ”redirectable” flagCongested border router → DODAG Redirection Solicitation (DRS)
6
Network link
Internet
54
BR2
BR1
3
12
Control messagesBR1 instance BR2 instance
BR1→5 nodesBR2→1 node
Figure 8: Redirection of node 4 from BR1 to BR2
Brandon Foubert Sharing is caring 7 / 17
Load balancing
Border router redundancy → static (i.e. non-adaptative) load balancing⇒ RPL + explicit redirection:
Multiple RPL instances → border router differentiationColocated networks → nodes set ”redirectable” flagCongested border router → DODAG Redirection Solicitation (DRS)
DR
S
DIS
6
Network link
Internet
54
BR2
BR1
3
12
Control messagesBR1 instance BR2 instance
Figure 8: Redirection of node 4 from BR1 to BR2
Brandon Foubert Sharing is caring 7 / 17
Load balancing
Border router redundancy → static (i.e. non-adaptative) load balancing⇒ RPL + explicit redirection:
Multiple RPL instances → border router differentiationColocated networks → nodes set ”redirectable” flagCongested border router → DODAG Redirection Solicitation (DRS)
DIO
6
Network link
Internet
54
BR2
BR1
3
12
Control messagesBR1 instance BR2 instance
Figure 8: Redirection of node 4 from BR1 to BR2
Brandon Foubert Sharing is caring 7 / 17
Multiple IPv6 prefixes
Considered scenario → multiple distinct IPv6 prefixes⇒ RPL + IPv6 Network Prefix Translation (NPT) [WB11]⇒ Prefixes sharing → backup routes → multi-homing
Prefix A
Network link
BR1 Internet
DODAG link Packet path
Figure 9: Address translation upon border router packet forwarding
[WB11] M. Wasserman and F. Baker. IPv6-to-IPv6 Network Prefix Translation. RFC 6296. June 2011
Brandon Foubert Sharing is caring 8 / 17
Multiple IPv6 prefixes
Considered scenario → multiple distinct IPv6 prefixes⇒ RPL + IPv6 Network Prefix Translation (NPT) [WB11]⇒ Prefixes sharing → backup routes → multi-homing
Prefix A
Network link
BR1 Internet
DODAG link Packet path
Figure 9: Address translation upon border router packet forwarding
[WB11] M. Wasserman and F. Baker. IPv6-to-IPv6 Network Prefix Translation. RFC 6296. June 2011
Brandon Foubert Sharing is caring 8 / 17
Multiple IPv6 prefixes
Considered scenario → multiple distinct IPv6 prefixes⇒ RPL + IPv6 Network Prefix Translation (NPT) [WB11]⇒ Prefixes sharing → backup routes → multi-homing
Prefix A
Network link
BR1 Internet
DODAG link Packet path
Figure 9: Address translation upon border router packet forwarding
[WB11] M. Wasserman and F. Baker. IPv6-to-IPv6 Network Prefix Translation. RFC 6296. June 2011
Brandon Foubert Sharing is caring 8 / 17
Multiple IPv6 prefixes
Considered scenario → multiple distinct IPv6 prefixes⇒ RPL + IPv6 Network Prefix Translation (NPT) [WB11]⇒ Prefixes sharing → backup routes → multi-homing
Prefix B
Network link
BR1 Internet
DODAG link Packet path
Figure 9: Address translation upon border router packet forwarding
[WB11] M. Wasserman and F. Baker. IPv6-to-IPv6 Network Prefix Translation. RFC 6296. June 2011
Brandon Foubert Sharing is caring 8 / 17
Multiple IPv6 prefixes
Considered scenario → multiple distinct IPv6 prefixes⇒ RPL + IPv6 Network Prefix Translation (NPT) [WB11]⇒ Prefixes sharing → backup routes → multi-homing
Network link
BR1 Internet
DODAG link Packet path
Prefix B
Figure 9: Address translation upon border router packet forwarding
[WB11] M. Wasserman and F. Baker. IPv6-to-IPv6 Network Prefix Translation. RFC 6296. June 2011
Brandon Foubert Sharing is caring 8 / 17
Outline
1 Scientific context
2 Contribution
3 ExperimentationExperimental setupTopologiesBandwidth repartitionEnd-to-end packet error rateNumber of one-hop transmissionsEnergy consumption
4 Conclusion
FIT/IoT-LAB
Figure 10: Strasbourg testbed
Brandon Foubert Sharing is caring 9 / 17
Hardware
Figure 11: M3 Open Node
Brandon Foubert Sharing is caring 10 / 17
Experimental setup
Contiki OS 3.x → Contiki RPLFIT/IoT-LAB testbed, M3 nodes
ParametersIEEE 802.15.4 CSMA/CAno radio duty cycle mecanism1 UDP packet per secondsub-DODAG size threshold as congestion trigger
Scenario2 border routers & 8 traffic generating nodesBorder router 53 wakes up 60s after border router 18100 experiments of 1h each
Brandon Foubert Sharing is caring 11 / 17
Topologies
(a) RPL topology (b) RPL-NPT-LB topology
Figure 12: Cumulative final DODAGs from all experiments(the thicker a link is, the more frequently it appears)
Brandon Foubert Sharing is caring 12 / 17
Bandwidth repartition
0 400 800 1200 1600 2400 2800 3200
Time (s)
0
80
160
240
320
400
480
2000
(1 second intervals)
RPL-NPT-LB BR18 RPL-NPT-LB BR53RPL BR18Bandwidth(Bytes)
○≈ 460B/s
○≈ 480B/s
t1 t2 t3
(cumulative)
Figure 13: Better division of the traffic load between border routers
Brandon Foubert Sharing is caring 13 / 17
End-to-end packet error rate
26 30 34 38 48 54 56 58Node n°
12
10
8
6
4
2
0PE
R di
ffere
nce
(%)
Figure 14: End-to-end losses difference between RPL-NPT-LB and RPL (lower is better)
Brandon Foubert Sharing is caring 14 / 17
Number of one-hop transmissions
26 30 34 38 48 54 56 58Node n°
0
5000
10000
15000
20000
25000
30000
35000
40000Nu
mbe
r of t
rans
miss
ions
Figure 15: Number of transmissions (red is RPL — green is RPL-NPT-LB)
Brandon Foubert Sharing is caring 15 / 17
Energy consumption
26 30 34 38 48 54 56 58Node n°
3.0
2.5
2.0
1.5
1.0
0.5
0.0Co
nsum
ptio
n di
ffere
nce
(J)
Figure 16: Energy consumption difference between RPL-NPT-LB and RPL (lower is better)
Brandon Foubert Sharing is caring 16 / 17
Outline
1 Scientific context
2 Contribution
3 Experimentation
4 Conclusion
Conclusion
IoT and RPL → single point of failure (border router)Colocated networks → cooperation for redundancy
Future workExperiment with larger and random network layoutsDifferent congested mode triggersPrecise assessment before redirection (e.g. link quality)
Brandon Foubert Sharing is caring 17 / 17
Conclusion
IoT and RPL → single point of failure (border router)Colocated networks → cooperation for redundancy
Future workExperiment with larger and random network layoutsDifferent congested mode triggersPrecise assessment before redirection (e.g. link quality)
Brandon Foubert Sharing is caring 17 / 17
RPL cooperation experiments using FIT IoT-LAB testbedContact: brandon.foubert@inria.fr
Thank you for your attention!Any questions?
[FM19] Brandon Foubert and Julien Montavont. “Sharing is caring: a cooperation scheme for RPL network resilience andefficiency”. In: ISCC 2019 - 24th Symposium on Computers and Communications. June 2019
Control messages
DIO DAO DIS DRS0
20
40
60
80
100
120
140
160
180Number of control messages
RPLRPL_NTP_LB
Figure 17: Transmission number of control messages
Experimental parameters
MAC layer IEEE 802.15.4 CSMA/CAMAC acknowledgments Enabled
MAC Tx queue size 1 packetRDC mechanism No RDC (NULLRDC)
Traffic type UDP packetsTraffic rate 1 packet per secondTx power 3 dBm
Rx power threshold -60 dBmMotes used 10 M3 open nodeRPL mode Non-storing
RPL OF MRHOF ETXCongested mode trigger Sub-DODAG size threshold
Repartition of transmission state
26 30 34 38 48 54 56 58Node n°
0
5000
10000
15000
20000
25000
30000
35000
40000Nu
mbe
r of t
rans
miss
ions
Tx ok No ack Ack tx Ack collision Queue drop
Figure 18: Repartition of transmission state (left RPL — right RPL-NPT-LB)