You are on page 1of 43

Opportunistic Networking :

Data Forwarding in
disconnected mobile ad
hoc networks
Abstract

 Opportunistic networks are on of the most


interesting evolution of MANETs
 Opportunistic network mobile nodes are
enabled to communication with each node
 Routes are built dynamically

 messageห are en route between sender aก


destination
 Node can opportunistically be used as next
hop
 This method is to bring message closer to the
final destination
Introduction

 Research on multihop ad hoc network has a


focused on a number of application
environment
 Originally conceived for military application,
and aimed at improving battlefield
communication
 Two main evolutions of multihop ad hoc
network are mesh network and
opportunistic network
 Opportunistic network no assumption is
made with regard to the existence of
complete path between two nodes wishing
to communication
 Concept of opportunistic network come from
Taxonomy of routing/forwarding
techniques for Opportunistic Network
Delay-tolerant network

 Delay-tolerant network (DTN) architecture


consists of
 DTN Region
 DTN Gateway
 Use protocol stack that best suit the
particular infrastructure
 DTN node overlay protocol is added on top of
the traditional transport layer


Delay-tolerant network
 Architecture

Region A - Internet dat


a

{B, R2}
dat
{A, R2}
a {A, R1}

dat
{D, R4}
a
Region B – Sensor network
{C, R4}

UserHost
{B, R3} {C, R3}
{A, UserHost}
dat
a

6
DTN gateway
Source : http://www-net.cs.umass.edu/~shyang/presentation_slides/DTN.ppt
Delay-tolerant network

Example of delay­tolerant network
Opportunistic Network

Oppo rtunistic  ne two rking


Realistic Case Studies

 PSNs in the Haggle Project


 Wildlife Monitoring :
 Zebranet
 SWIM
 Opportunistic Network for Developing Areas
 DarkNet
 Saami Network Connectivity (SNC)
Realistic Case Studies

 Research on opportunistic network is


devoting particular attention to realistic
case study
 On of the basic component realistic case
studies are mobility models
 Design of efficient forwarding algorithm as
well as to perform realistic simulation
 Researchers are also implementing a number
of real-application scenarios in
opportunistic network
 For example, with wildlife tracking
application

Haggle Project

 It targets solutions for communication in


opportunistic network
 Funded by European Commission in the
framework of the FET-SAC
 Studying the properties on Pocket Switched
Networks (PSNs)
 Pair-wise modeling contacts between devices
Wildlife Monitoring : ZebraNet

 Biologists want to track animals


 Long-term
 Over long distances
 Questions:
 Interactions within a species?
 Interactions between species?
 Impact of human development?

Source : http://www.princeton.edu/~mrm/zebranet.html
Wildlife Monitoring : ZebraNet

Source : http://www.princeton.edu/~mrm/ZNetASPLOS.pdf
ZebraNet Protocols

 Two peer-to-peer protocols evaluated here


 Flooding: Send to everyone found in peer
discovery.
 History-Based: After peer discovery, choose
at most one peer to send to per discovery
period: the one with best past history of
delivering data to base.
 Simulation results show that both protocols
outperform the directly protocol
 History-based protocol outperforms flooding
in term of bandwidth and energy
consumption

Wildlife Monitoring : SWIM

 In the Shared Wireless Infostation Model


(SWIM) whales are monitored.
 Data is replicated and diffused (similar to the
flooding protocol in ZebraNet)
 Both whale-to-whale and whale-to-base-
station communications are allowed
 No experimental result are actually available
but simulation results are quite realistic
since simulation parameter set according
to studies conducted by biologist whales’
real habits

Opportunistic Network for Developing
Areas : Darknet
 It aims to provide a low-cost Internet
Connectivity to rural villages in India
 Kiosks are built up and equipped digital
storage and short-range wireless
communication
 MAPs exchange data with the kiosks
wirelessly
 MAPs can download/upload data to the
Internet when passing by APs in a nearby
town
 It supports Internet/Intranet messaging,
distribution and collection information
Opportunistic Network for Developing
Areas : SNC
 It aims to provide network connectivity to the
nomadic Saami population of the reindeer
herders
 Providing network connectivity is a mean to
protect their habits, culture and tradition
 In its initial stage, it focused on providing
email, file transfer, and cached web
services
 It should finally be noted that the SNC project
focuses on a pure DNT architecture

Opportunistic Routing/Forwarding
Techniques
 Routing without Infrastructure
 Dissemination-Based Routing
 Context-Based Routing
 Routing with Infrastructure
 Routing Based on Fixed Infrastructure
 Routing Based on Mobile Infrastructure
(Carrier-Based Routing)


Dissemination-Based Routing

 Based on data dissemination perform


delivery a message to destination by
diffusing it all over the network.
 No knowledge of a possible path to
destination and nor of appropriate of next-
hop
 Works well on mobile network

 Limits message delay but is resource hungry.

 May lead to network congestion


Dissemination-Based Routing

 Such as
 Epidemic Routing Protocol
 MV Routing Protocol
 Network Coding-Based Routing Protocol

Epidemic Routing protocol

 Goal is to deliver messages with high


probability even when there is never a fully
connected path.
 Give a message copy to every node
encountered
 essentially: flooding in a disconnected
context

Source : http://www.cise.ufl.edu/~helmy/cis6930-09/Epidemic-and-Ferry.ppt
Mv routing protocol

 A further step beyond epidemic routing


 Messages exchanged during pair-wise
contacts as in epidemic routing but use
more sophisticated method to select
messages to forward to a
 Deliver probability is relied on recent-past
observation of both meeting between
nodes and the visits of nodes to
geographical location
 A similar approach is followed in PROPHET
routing protocol

Network coding-based protocol

 Takes an original approach to limit message


flooding
 Outperform flooding as it is able to deliver
the same information with a fewer number
of messages infected into the network
 See example on next slide


Example  o f ne two rk­c o ding e ffic ie nc y
Context-based routing

 Exploits more information about the context


so as to identify suitable next hops towards
the destinations (e.g., the home address of
a user)
 Reduce messages duplication

 Tend to increase the delay that each


message experience during delivery
 Due to errors and inaccuracies in selecting
the best relays
Context-based routing (contd.)

 Nodes maintain a state in order to keep track


of the utility values
 Need storage capacity for both state and
messages
 Cost to hold and update the state at each
node (overhead)
Context-aware routing (CAR) Protocol

 Each node is in charge of producing delivery


probabilities towards each destination host
 Delivery probabilities are exchanged so that
node compute the best carrier for
destination node based on the nodes’
context
 When the best carrier receives a message, it
stores it in a buffer and forwards it to the
destination node when met
 CAR provides a framework for computing
next hop based on the multiattribute utility
theory

Mobyspace : mobility pattern space
routing protocol
 The nodes’ mobility pattern is the context
information used for routing
 The protocol builds up a high dimensional
Euclidean space
 Each axis represents a possible contact
between a couple of nodes
 The distance along an axis measures the
probability of that contact to occur
Example scenario

Source : http://jeremie.leguay.free.fr/lip6/files/poster_sigcomm05.pdf
Node mobility pattern

Source : http://jeremie.leguay.free.fr/lip6/files/poster_sigcomm05.pdf
Routing with infrastructure

 Routing based on fixed infrastructure


 Routing based on mobile infrastructure
(carrier-based routing)
Routing based on fixed infrastructure

 A source node delivery a message keeps it


until it reach a base station belonging to
the infrastructure
 Then forwards the message to it

 Base station are gateways towards less


challenged networks (e.g. connected to a
LAN)
 The goal of an opportunistic routing
algorithm is to delivery messages to the
gateways, which are able to find the
destination more easily
Two variations of the protocol

1.First, only node-to-base-station


communications
 Work as described above
 messages experience high delays
 The example of this approach is the
Infostation model
Two variations of the protocol (contd.)

2.Second, both node-to-base-station and node-


to-node communications
A node to send a message to a destination
node delivers the message to the base
station directly, if within communication
range
 Otherwise, it delivers the message to a near
node that will forward it to the base station
 The example of this approach is the Shared
Wireless Infostation model (SWIM)
As result from the above examples

 Fixed base station play a passive role in the


opportunistic forwarding strategy because
act as information sinks
 benefits by running an opportunistic routing
algorithm at base stations
 Base stations can collect the messages sent
by visiting nodes
 Then wait for the destination nodes reach to
forward and the stored messages to them
Routing based on mobile infrastructure
(carrier-based routing)

 Nodes of the infrastructure are mobile data


collectors
 They move around in the network area and
gather messages from the nodes they pass
by
 These special nodes are referred to as
carriers, supports, forwarders, MULEs, or
even ferries
 They are entities responsible for messages
delivery
Routing based on mobile infrastructure
(carrier-based routing) contd.

 Node-to-carrier communications
 help increasing connectivity in sparse
network and guaranteeing that isolated
nodes be reached
 delivery of messages is accomplished
 Both by carriers and ordinary nodes
 Both node-to-node and node-to-carrier

The data-mule system

 Focus on data retrieval from sparse wireless


sensor networks
 Consist of a three-tire architecture:
 The lower level : sensor nodes periodically
perform data sampling from the surrounding
environment
 The middle level : mobile agents (MULEs)
move around in the area to gather their
data
 The upper level : a set of wired Aps and data
repositiories which receive information from
the MULEs and connected to a central data
warehouse
The message-ferrying approach

 Extra mobile nodes (message ferries) are


exploited to offer a message relaying
service
 Nodes move around in the network

 They collect messages from source nodes


Message collection happen in
two ways:
 Node-initiated message ferrying :
 the ferry node moves around a predefined and
known path
 Each node has knowledge of the path by
ferries
 node moves to meet ferries when it has data
to deliver
 Ferry-initiated message ferrying :
 source node sends a ServiceRequest and
current position to the ferry
 the ferry changes trajectory to meet the
source node

Conclusion

 Taxonomy of routing/forwarding techniques


for opportunistic networks : without and
with infrastructure
 Interesting is how to design multitier
opportunistic networks
 The data MULEs and message-ferrying
architectures are the most promising
approach
 In the data MULEs approach, lower level
nodes exploit the higher level and mobile
device
(the MULEs)
Future trends

 Each level of the infrastructure is an


opportunistic network in which nodes may
exploit routing algorithms to communicate
 and may rely on the upper levels of the
infrastructure to reach nodes are too far
away
Future trends (contd.)

 For example, lower level : devices (e.g. PDA,


smart phone)
 An opportunistic routing algorithm make
devices to communicate with each other
 Higher level : city-bus network might be used
, bus act as MULEs
 City-bus network might exploit further level
such as mesh network
 Opportunistic network might represent
a fundamental building block for the next-
generation internet

You might also like