The document discusses the performance evaluation of the Binary Spray and Wait opportunistic network (OppNet) routing protocol in the context of emergency scenarios. It motivates the use of OppNets for emergency communications when traditional infrastructure is unavailable. Simulation results show that the Binary Spray and Wait protocol achieves good delivery probability and latency. The optimal buffer size is found to be 30MB, and speedier nodes like cars can carry larger message sizes like images but not videos. Future work includes testing the protocol on real emergency scenario data and smartphones.

1. Mazlan Abbas (MIMOS Berhad), Nur Husna Md Yusof
(Universiti Teknologi Malaysia) and Norsheila Fisal (Universiti
Teknologi Malaysia)
Performance Evaluation of Binary Spray and
Wait OppNet Protocol in the Context of
Emergency Scenario
PerNEM'2013,'San'Diego,'USA'
Contact: mazlan.abbas@mimos.my
@mazlan_abbas
h7ps://www.facebook.com/drmazlanabbas'

2. Outline
• Motivation
• Routing Challenges in Emergency
Scenario
• Simulation Result
• Summary
• Future Work

3. Issues of Communications During Disasters and Emergency
Terrorist'A7ack' Hurricane'
Tsunami' Earthquake'
Lack'of'
adequate'
communic
aLons'

4. Opportunistic Networks (OppNets)
To'enable'communicaLon'between'source'and'desLnaLon'
without'the'support'of'a'fixed'network'infrastructure'

5. Smartphones – The Enabler
Camera'GPS'
WiFi'
Big'Storage'
CPU'Power'

6. History of Delay Tolerant Network
• Interplanetary'Internet'(IPN)'is'a'NASA'research'project'led'by'Vint'Cerf'in'
1998.
• The'basic'idea'is'to'try'to'make'data'communicaLons'in'space/'between'
planets.
• IPN'became'the'most'fundamental'basis'for'DTN'architecture'and'protocol'
suite.'
• The'Interplanetary'Internet'is'a'disconnected,'storeand'forward'
network'of'Internets 'based'on'a'wireless'backbone'with'huge'
delays'(The'delay'in'sending'or'receiving'data'from'Mars'takes'
between'3.5'to'20'minutes'at'the'speed'of'light)'and'error'prone'
links
• Failing'of'IP/TCP'in'space'missions
– Endtoend'path'exist
– Small'delays'

7. Delay Tolerant Networking (DTN)
• DTN'is'a'set'of'protocols'that'
act'together'to'enable'a'
standardized'method'of'
performing'storecarryand
forward'communicaLons.
• CharacterisLcs'of'DTN:
i. Intermi7ent'connecLvity
– No'endtoend'path'between'source'
and'desLnaLon
ii. Long'variable'delay
– Long'propagaLon'delays'between'
nodes'
A'
B'
B'
C'
C' D'
Source'
Store'
Carry'
Forward'
Store'
Carry'
Forward'
Delay'Tolerant'Network'(DTN)'='Mobile'OpportunisLc'Network'(OppNet)'

8. 8'
3G'
WiFi'
WiFi'
WiFi'
3G'
3G'
3G'Base''
staLon'
OppNet'in'Emergency'Response'Scenario'
Public Safety

9. 9'
3G'
WiFi'
WiFi'
WiFi'
3G'
3G'
Base''
StaLon'down'
X' X'
X'X'WiFi' WiFi'
OppNet'in'Emergency'Response'Scenario'
Public Safety Smartphones'(Nodes)'can'be'
carried'by' Pedestrians 'or'
Vehicles '
Send'“SOS”'messages'
Send'photos'of'vicLms'or'self'
Can$we$send$videos?$
What$kind$of$file$size?$

10. 10'
3G'
WiFi'
WiFi'
WiFi'
3G'
3G'
Base''
StaLon'down'
X' X'
X'X'WiFi' WiFi'
Internet'
Internet'
Ability'to'Connect'to'Internet'at'
Remote'Ends'
Public Safety
Ques7on:$Can$smartphones$help$us$
during$Emergency$Situa7on?$

11. 11'
3G'
WiFi'
WiFi'
WiFi'
3G'
3G'
Base''
StaLon'down'
X' X'
X'X'WiFi' WiFi'
Internet'
Internet'
DTN'Gateway'
DTN'Gateway'

12. C1'
C6'
C3'
C2'
C3'
C5'
Node'Mobility'
Source'
DesLnaLon'
Routing Challenges
Example:$Disaster'relief'efforts,'mining'operaLons,'health'campaigns'
In'emergency'situaLons,'enLLes'with'any'sensing'capabiliLes''such'as'cellphones'with'GPS'or'desktops'
equipped'with'surveillance'cameras,'can'be'especially'valuable'for'the'OppNet.'
C4'

13. N1'
Factors That Impact Performance
N2'
Mobility'Pa7ern'
Node'Speed'
Type'of'CommunicaLon'
Transmission'Range'
Buffer'Size' Ba7ery'Life'TimetoLive'Message'Size'
RouLng'Mechanism'
Number'of'Nodes'
Size'of'Area'

14. 1
4
Routing Protocols – Related Works (1)
• The Direct Delivery does not start any further transactions after
exchanging the deliverable messages since it will send messages
only if it is in contact with the final recipient.
• While in the First Contact routing, it sends as many messages to
the other node as it has time; it removes the local copy of the
message after a successful transfer. This results in only a single
copy of every message in the network.
• Epidemic routing [Vahdat et al 2000] spreads an unlimited number
of message copies by having nodes replicate them to all other
nodes they connect. This includes the messages they create and
the messages they have received from other nodes.
14'

15. 1
5
Routing Protocols – Related Works (1)
• Predictive protocols such as PRoPHET [Lindgren et al 2004] use past
encounters of nodes to predict their future suitability to deliver
messages to a certain target. It uses a metric called delivery
predictability that is based upon how often two nodes meet each other.
The more frequently and the more recently these nodes have met, the
better a forwarder one is for messages directed to the other.
• While PRoPHET checks if another node is more likely to meet the final
recipient, MaxProp [Burgess et al 2006] uses Dijkstra s algorithm to
calculate whole paths from node to node using the meeting
probabilities.
• The Spray and Wait [Spyropoulos et al 2005] works a bit like the
Epidemic but it restricts the amount of copies that are spread in the
network. Letting each created message to replicate only a certain
amount of times. A node that has more than one copy of the message
left, can give either a one copy to another node (the normal mode) or
half of the copies (the binary mode).
15'

16. 1
6
Spray & Wait - A Better Choice
16'
Spray'&'Wait'–'High'delivery'probability' Spray'&'Wait'–'Low'overhead'(First'contact'don’t'have'
good'delivery'probability)'
Spray'&'Wait'–'Low'latency'
[Source:'“performance'of'Challenged'Internet'of'Things'in'Emergency'Response”,'M.'Abbas'and'N.H.Md.'Yusuf,'Technical'Report,'2012]'
Shortest$Path$MapDBased$Movement$Model$–'Assuming'in'
most'emergency'situaLon,'people'and'vehicle'will'move'
along'the'roads.''
!

17. 17'
Spray and Wait Routing Protocol
• In Spray and Wait, message is delivered in two phases;
the spray phase and the wait phase.
• In the spray phase, source node spread a small number
of copies to only a few relays. A node that has more than
one copy of the message left can give either a copy to
another node (the normal mode) or half of the copies (the
binary mode) and keeps the rest to itself.
• In the wait phase, if the node has only a single copy of
the message left, it is directly transmitted only to the
destination.
• We use Spray-and-Wait in binary mode: a node carrying
k copies of a message forwards k/2 of them to the next
nodes it meets until the k =1. Then, a node waits till it
meets the destination.

18. 18'
Performance Metrics
• Delivery probability: It is a ratio between the
number of messages arrives at destination and the
number of messages sent.
• Message dropped: It is the number of messages
dropped from nodes' buffers during transmission.
Messages are dropped once the buffer is full.
• Latency average: The latency average is an
average time taken for a message to reach
destination.
• Hop count average: It is an average number of
hops between source and destination nodes.

19. 1
9
SimulaLon'tool:'ONE'(OpportunisLc'Networking'Environment)'
Helsinki'Downtown'Map'(4.5km'x'3.4'km)'
Assumptions
19'

20. 20'
Parameter' Value'
SimulaLon'Time'(s)' 43200'
SimulaLon'Area'(sq.m)' 4500x3400'
RouLng'Protocol' Binary'Spray'and'Wait'
Number'of'Copies' 6'
Mobility'Model' Shortest'Path'Map'Based'Movement'
Transmission'Range'(m)' 10'
Transmit'Speed'(Mb/s)' 10'
TTL'(minutes)' 300'
Main Simulation Parameters

21. 21'
Experiment' Group' Number'of'Nodes' Speed'(m/s)'
1' Pedestrians' 20' 0.51.5''
2' Cars' 20' 2.7'–'13.9'
3' Pedestrians' 100' 0.5'–'1.5'
4' Cars' 100' 2.7'–'13.9''
Experiments – Varying Buffer Size
Parameter' Value'
Buffer'size'(MB)' 5100'
Message'size'(B)' 500k'–'1M'
Message'generaLon' Every'30'seconds'

22. 22'
Varying the Buffer Size
Delivery'Probability'vs.'Buffer'Size''
Impact'on'Delivery'Probability'
Speed'of'nodes'does'not'
ma7er'with'buffer'size'but'
number'of'nodes'does'
ma7er'
Higher'loss'of'messages'when'buffer'size'
is'too'small.'OpLmum'at'30'MB.'
Messages'eventually'have'to'be'dropped'
if'it'exceeded'the'TTL.'

23. 23'
Varying the Buffer Size
Number'of'Messages'Dropped'vs.'Buffer'Size''
Impact'on'Number'of'Messages'Dropped'
Beyond'30'MB'buffer'size,'the'number'of'messages'being'dropped'remained'the'same.'

24. 24'
Varying the Buffer Size
Latency'Average'vs.'Buffer'Size''
Impact'on'Latency'Average'
Smaller'number'of'
nodes'has'higher'
latency.'

25. 25'
Varying the Buffer Size
Hop'Count'vs.'Buffer'Size''
Impact'on'Average'Hop'Count'

26. 26'
Experiment' Group' Number'of'Nodes' Speed'(m/s)'
1' Pedestrians' 20' 0.51.5''
2' Cars' 20' 2.7'–'13.9'
3' Pedestrians' 100' 0.5'–'1.5'
4' Cars' 100' 2.7'–'13.9''
Experiments – Varying Message Size
Parameter' Value'
Message'size'' 500'kB''to''5'MB'
Buffer'size' 30'MB'

27. 27'
Varying the Message Size
Delivery'Probability'vs'Message'Size'
Impact'on'Delivery'Probability'
Slower'nodes'able'to'carry'bigger'message'sizes'

28. 28'
Varying the Message Size
Average'Latency'vs'Message'Size'
Impact'on'Average'Latency'

29. 29'
Summary
• Opportunistic Networks (OppNets) are very
useful in the context of emergency scenarios
• Binary Spray and Wait Protocol is one good
option for routing
• Smartphones seems to be a good potential
candidate communications tool in emergency
scenarios
• Speedier nodes (cars) require smaller Message
size (images rather than videos).

30. Future Work
• Requires actual datasets (emergency scenarios)
for mobility model
• Mixed mobility scenarios (pedestrians plus
vehicles)
• Find better battery efficient methods
• Further refinements to Spray & Wait Protocol
with other parameters such as Contact Time etc.
• Implementation on smartphones – e.g. WiFi
Direct 802.11ac or Bluetooth

31. THANK YOU
Contact: mazlan.abbas@mimos.my
@mazlan_abbas
h7ps://www.facebook.com/drmazlanabbas'