Probability-based Controlled Flooding in Opportunistic Networks

Sanjay K. Dhurandher

, Satya Jyoti Borah

, Mohammad S. Obaidat

, Deepak Kr. Sharma

Sahil Gupta

and Bikash Baruah

CAITFS, Divison of Information Technology, Netaji Subash Institute of Technology, University of Delhi, New Delhi, India

Dept. of Comp. Sc. and Eng., Monmouth University Monmouth, NJ, U.S.A.

Department of Computer Science and Engineering, North Eastern Regional Institute of Science and Technology, Deemed

University, Itanagar, India

Keywords: Opportunistic Networks, Opportunistic Network Simulator (ONE) Environment, Epidemic, Spray and Wait,

Prophet, Routing Protocol.

Abstract: Opportunistic networks (Oppnet) are challenged networks in present wireless communication scenario. These

networks are mainly applied to situations where a persistent end–to-end path between the source and the

destination does not exist. Delay/Disruption Tolerant Networking (DTN) is mostly used to solve this end-to-

end path problem in such networks. Many routing protocols have been proposed in literature that consider

various performance metrics such as delivery delay, packet delivery rate, hop count, among others. In this

paper, a new routing protocol named as PRoWait has been designed which can overcome the shortcomings

of the already existing protocols in Oppnets. The proposed protocol also incorporates the merits of existing

protocol so that it can be reliable and efficient for the communication of pedestrians with handheld devices.

Simulation results obtained for the proposed scheme show better performance as compared to the Porphet,

Spray and Wait and Epidemic routing protocols in terms of packets delivery probability, overhead ratio, and

hop count performance metrics.

1 INTRODUCTION

An Ad hoc network (Royer, 1999) is a peer to peer

communication network. It uses wireless/Wi-Fi tech-

nology and the communication takes place without

any access point. It is an infrastructure less, cost effi-

cient, quicker to setup, and works as an extension to

existing networks. In this chain, there is one more

specific network where the nodes are mobile in na-

ture. This network is called as mobile Ad hoc network

[MANET] (Toh, 2002). The main aim of MANETs is

to establish a synchronous communication between

two or more nodes. In a multi-hop communication en-

vironment for MANET, every node is assumed to be

helpful to each other. They agree to contribute and

route any traffic within the network. However, they

have some limitations such as minimal security

against unwanted incoming connections, which

makes it easy for attackers.

Opportunistic networks (Pelusi et all, 2006) are

the recent evaluations of MANETs. They have many

similarities and differences to the MANETs. MA-

NETs follow the synchronous mode of communica-

tion while Oppnets follow the asynchronous mode of

communication. In MANETs first a fixed path is es-

tablished between the source and destination and then

the message passing takes place. But, in Oppnets no

previous assumption is made regarding the existence

of a complete path between the two nodes that wish

to communicate with each other (Huang et al., 2008).

Source and destination nodes might not be in the same

network or within range of each other at the same

time. However, Oppnet provides a platform for nodes

to communicate in the aforementioned type of chal-

lenged network scenarios. Packets are often buffered

in the network waiting for a path towards the destina-

tion to be available. Due to this, additional delay gets

incorporated in message delivery. However, there are

various applications which can tolerate longer delays

such as e-mailing. Again frequent connections and

disconnections between nodes may degrade the per-

formance of a number of applications in Oppnets.

K. Dhurandher S., Jyoti Borah S., S. Obaidat M., Sharma D., Gupta S. and Baruah B..

Probability-based Controlled Flooding in Opportunistic Networks.

DOI: 10.5220/0005580300030008

In Proceedings of the 12th International Conference on Wireless Information Networks and Systems (WINSYS-2015), pages 3-8

ISBN: 978-989-758-119-9

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

Routing and forwarding of packets in Opnets is a

very difficult task. It is completely different from tra-

ditional network routing techniques. Here, the main

objectives are to expect a reliable delivery of packets

even in the absence of permanent path between the

sender and the receiver. As the nodes are mobile in

nature in this kind of network, one cannot predict to-

tally about the routes for packets transmission. Hence,

conventional routing protocols do not have any role

to play in these networks (Pelusi et all, 2006).

Routing is mainly based upon three things

(Dhurandher et al., 2008). The first is a node’s coop-

eration that is its willingness in the routing and for-

warding. The second is nodes mobility which can be

utilized to forward a packet from one portion of the

network to the other. The third is store-carry-forward

method. This method ensures that the packet copy re-

mains saved in the buffer of the nodes until it meets

the next forwarding node or the destination node. It

also requires proper buffer management when buffer

gets full. Hence, there is always a need for proper

routing protocol that minimizes the delivery delay,

buffering space and maximize delivery ratio.

In this work, a new protocol named as PRoWait

has been proposed which is based on the concept of

earlier work of Spray and Wait and Prophet routing

protocols. It integrates the working principle of the

two aforementioned protocols together in order to

achieve a better packet delivery ratio, latency and hop

count. This paper does not address the acknowledge-

ment of packets and security issues in Oppnets. The

overall scenario is simulated with the help of ONE

simulator (Keranen, 2008) and respective results were

recorded.

The rest of the paper is organized as follows.

Section II gives a brief overview of existing protocols

that have been used in this work. Section III presents

the proposed protocol in detail. Section IV describes

the simulation scenarios and Simulation results with

various graphs. Section V summarizes this work.

2 RELATED WORK

So far many routing protocols have been designed in

the past in Oppnets such as Epidemic (Vahdat, 2000),

Spray and Wait (Spyropoulos et al., 2005), Binary

Spray and Wait (Spyropoulos et al., 2005), Prophet

(Lindgren et al., 2003), HBPR (Dhurandher et al.,

2013) (Dhurandher et al, 2014), GAER (Dhurandher

et al, 2014), among others. The main objective of these

protocols is to achieve maximum successful delivery

of packets and minimize the delay. In this section, a

brief discussion of some of these routing protocols is

presented.

A. Epidemic Routing Protocol

The Epidemic routing protocol, as its name suggests is

similar to spreading/flooding like infectious diseases

in nature. It is a simple protocol which delivers most

of the packets successfully to its destination, but

requires large amount of bandwidth and buffer size.

When a pair of nodes that want to communicate come

into a communication range, the first node sends the

second node, a summary vector containing

information that uniquely identifies the packets it has

in its buffer. The second node also transfers to the first

node any packets it has which are not available with

the first node. Hence transmission of packets takes

place in both directions whenever two nodes meet

each other. In this way all the packets will ultimately

be distributed to every node, and finally every packet

will reach its destination as quickly as possible with a

very high probability.

The main approach of this protocol is to distribute

and deliver application packets to hosts, called

carriers, within connected portion of Oppnets. By

using this scheme, packets are spread quickly within

the connected portion of the network through node

mobility. At this point, the packets get spread to an

additional group of nodes. Through such transitive

transmission of data, packets have a high probability

of eventually reaching their destination. This scheme

requires large amount of bandwidth and buffer space.

Moreover, a large number of multiple copies of the

same packets get generated throughout the network

which leads to network congestion.

B. Spray and Wait Routing Protocol

This scheme consists of the following two phases

(Spyropoulos et al., 2005):

 Spray phase: The packet originating at the

source node having X copies initially spread/for-

warded the copies by source or possibly other

intermediate nodes receiving a copy of X dis-

tinct relays.

 Wait phase: During spraying phase if no desti-

nation is found then each of the X nodes carrying

packets performs direct transmission.

This protocol uses the concept of Epidemic routing

due to its speediness and simplicity for packets

spreading and direct transmission. The scheme starts

by spreading numbers of packets in a manner that is

similar to the routing. By doing so, enough copies of

the same packet get generated in the network and

eventually at least one of them will find the

WINSYS2015-InternationalConferenceonWirelessInformationNetworksandSystems

destination as quickly as possible (with high

probability). In this scheme, the node stops spreading

packets when it has only one copy with it. The node

which is carrying a single copy performs the direct

transmission. Due to this mechanism Spray and Wait

could be viewed as single and multi-copy schemes.

Hence, its performance is quite better as compared to

Epidemic routing and other single and multi-copy

schemes in terms of number of transmissions and

delivery delay.

C. Prophet Routing Protocol

In the Probabilistic Routing Protocol using History of

Encounters and Transitivity) [10], if a node visited a

location many times in the past then the chance of

visiting same location by the same node is more. In

this scheme before sending packets a probabilistic

metric called delivery predictability is created, say P

(a;b)

ϵ [0 ; 1], at every source node “a” for each known

destination “b”. This delivery predictability reflects

how likely a node will be able to deliver a message to

the destination. When a node “a” comes in contact

with node “b” then node “a” will transfer the packet

to node “b” if and only if the delivery predictability

of the node “b” is higher than that of node “a”.

3 PROPOSED SCHEME:

PRoWAIT

In this section, the PRoWait protocol is discussed in

detail. It is a hybrid forwarding strategy that combines

the advantages of Prophet and Spray and Wait proto-

cols.

A. Forwarding Strategies

Most of the routing protocols designed so far forward

a packet from one node to its neighbor based upon

lowest cost of the path towards the destination. Also,

the packet is sent to a single node instead of multiple

nodes due to high reliability of multipath

communication. However, in Oppnets things are

completely different. When a packet arrives at a node,

there might not be any available path to the

destination and the node has to buffer it in its storage

space. When the source node encounters another

node, a decision has to be made whether a particular

packet is to be transferred or not. To increase the

delivery probability it may also be required to

forward a packet to multiple nodes. But, these types

of decisions are not trivial to make.

When a node encounters another node with low

value of delivery predictability, one cannot predict

whether this node will encounter another node with

higher probability within a reasonable period of time.

Furthermore, selecting a particular node out of many

available nodes for transmitting a certain packet is

another problem. However, with distribution of large

number of packets to a large number of nodes, the

probability of delivering a packet to its destination

will increase. As a result more resources are required

which results in wasting system resources. Thus,

instead of giving a packet to many nodes, if only a

few nodes are selected, then lesser number of system

resources will get utilized. This will lower the

probability of delivering a packet and incur high

delay.

In PRoWait, we have chosen a rather simple

forwarding strategy–when two nodes meet each

other, a packet is transferred to the other node if the

delivery predictability of the destination of the packet

is higher at that node. The delivery predictability of

PRoWait is calculated from Prophet routing protocol.

Spraying the packets to the neighboring nodes is done

with the Spray and Wait protocol’s technique, while

Prophet’s delivery predictability is used in the

selection of a node as a next hop.

Figure 1: Algorithm with an illustrative example.

The source node initially has x copies of packet to be

sent. Any node A that has (x >1) copies of the packet

(source or relay) and encounters another node B (with

no copies), hands over to B [x/2] copies and keeps

[x/2] copies for itself. When B encounters another

node C (with no copies), it hands over to C [x/4]

copies and keeps [x/4] copies for itself. This process

Probability-basedControlledFloodinginOpportunisticNetworks

will continue until a node is left with only one copy

or reach the destination. If the destination is not

found, then each of the nodes carrying a single copy

of the packet performs direct transmission. Here

every node is provided with two variables: hop count

and probability. When two nodes come within the

range of each other, then a packet is passed from a

node having lower probability to a node having

higher probability, else no transmission occurs.

PRoWait creates neither too less nor too much copies

of the packets in the network. It tries to maintain

enough copies of the packets that ensure its successful

delivery with lesser delay to the destination. It also

tries to reduce the resource consumption by limiting

the amount of flooding done in comparison to the

Epidemic routing protocol.

B. Algorithm

The algorithm used in PRoWait is also depicted in

Fig.1. In this figure, node A is the source node that

has x copies of message M1 with it. It has two neigh-

bors, node B and node C. Both nodes do not have

message M1 with them. The value of delivery predict-

ability at node A, B, and C are 0.2, 0.3, and 0.1, re-

spectively. Thus, according to PRoWait, node B is se-

lected and node C is rejected as it has higher value of

delivery predictability than node A. Node A then

transfers the [x/2] copies of the message to node B

and keeps [x/2] copies with itself. This process of

message copy transfer goes on from one node to an-

other node until x becomes equal to 1 or the message

TTL expires. All nodes that have x = 1transfer the

message to the destination on having direct contact

with it.

4 SIMULATION AND RESULTS

We evaluated the performance of the proposed

scheme using simulation analysis. Simulation for the

PRoWait has been done on the Opportunistic

Network Environment (ONE) simulator. It is a Java

based simulation environment that is capable of:

• Generating node movement using different

movement models.

• Routing packets between nodes with various

Oppnet routing algorithms.

• Visualizing real time mobility and packets pass-

ing in its graphical user interface.

For Simulation analysis, we consider six groups

of mobile node. The pedestrian groups have 40 nodes

and one group (term) has 2 nodes each. Out of six

groups three groups are of pedestrian type with a

walking speed of 0.5 – 1.5 m/s. Another group,

term/period group, has a speed of 2.7 – 13.9 m/s and

final group is of city bus with a moving speed of 10 –

30 m/s. A TTL of 300 minutes is assigned to each

group for their messages. Group of pedestrians has a

buffer size of 50M each while the group of term has

50M buffer size. Transmission speed for pedestrians

is 2Mbps with a transmission range of 10m each and

for term it is 10Mbps with 1000m transmission range.

Each simulation is run for 20,000 second. A new mes-

sage of size 500Kb – 1Mb is created after every 25–

35 seconds. For movement of nodes Shortest path

map based movement model [14] is used with world

size of 4500 m x 3400 m.

The performance of PRoWait was evaluated and

compared to the Epidemic, Prophet, Spray and Wait

protocols by varying the buffer size and TTL. The

buffer size of each node in the scenario was changed

from 2MB to 15 MB and the resultant outputs were

observed. Further, evaluation was carried out by var-

ying the TTL of each message from 90 minutes to 210

minutes. The corresponding results were recorded,

analyzed, and discussed next.

Fig.2 shows the relation between the buffer size

and delivery probability. It can be observed from

Fig.2 that the delivery probability of packets for PRo-

Wait is found to be 0.4436 which is the second highest

among the four protocols plotted (Prophet = 0.2955,

Spray and Wait = 0.4522 and 0.2855 is for Epidemic).

Fig.3 shows the relation between buffer size and

mean overhead ratio. It is found that the mean average

overhead ratio for the proposed protocol is 8.703

which is the least among the four protocols plotted

(Spray and Wait = 11.65, Prophet = 52.136, Epidemic

= 78.021). Fig.4 depicts the relation between buffer

size and hop count. It can be observed from Fig.4 that

the mean value of hop count in PRoWait is 2.245,

which is lower than other three protocols (Prophet =

2.93, Spray and Wait = 2.35 and 3.86 is for Epidemic).

Figure 2: Comparison graph between Buffer size and De-

livery Probability.

WINSYS2015-InternationalConferenceonWirelessInformationNetworksandSystems

Figure 3: Comparison graph between Buffer size and Aver-

age Latency.

Fig.5 illustrates the relation between TTL and de-

livery probability of the four protocols. It is observed

that the mean delivery probability of packets for PRo-

Wait is 0.7492 which is the clear highest among the

four protocols plotted (Spray and Wait = 0.6536,

Prophet = 0.7115, and 0.5332 is for Epidemic). Due

to this, the number of packets successfully delivered

is quite high in PRoWait. Fig.6 is depicts the relation

between TTL and overhead ratio. It has been observed

that the mean overhead ratio for PRoWait is 14.05,

which is the lowest among the four protocol plotted

(Prophet = 45.42, Spray and Wait = 23.07 Sec, Epi-

demic = 181.133). Further, the ProWait performance

is 61.57% better than the Epidemic, 68.25% better

than Spray and Wait and Prophet, respectively in

terms of mean overhead ratio.

Figure 4: Comparison graph between Buffer size and Hop

count.

Figure 5: Comparison graph between TTL and Delivery

Probability.

Figure 6: Comparison graph between TTL and Average La-

tency.

Fig.7. shows the relation between TTL and hop count.

It is found that the mean value of hop count in

PRoWait is 2.07 which is again the lowest when

compared to the other three protocols (Spray and Wait

= 2.28, Prophet = 3.222 and 4.88 is for Epidemic).

Figure 7: Comparison graph between TTL and Hop Count.

Probability-basedControlledFloodinginOpportunisticNetworks

5 CONCLUSION

In this paper, a new routing protocol named as

PRoWait is designed to overcome some of the

shortcomings of the existing protocols in Oppnet. As

the area of routing in Oppnets is still under research

and much work is to be done in different aspects, the

main focus in this work is mainly to increase the

packet delivery ratio and lower delay and hop count.

The PRoWait is designed to overcome some of the

shortcomings of the existing protocols in Oppnet.

Through simulation analysis the performance of

PRoWait is evaluated and compared with Epidemic,

Spray and Wait, and Prophet protocols in terms of

delivery probability, overhead ratio, and average hop

count performance metrics. It has been observed that

PRoWait outperforms these protocols on the basis of

aforementioned performance metrics.

In future, efforts will be focused on increasing the

packet delivery ratio by adding some more

parameters and functions to the PRoWait scheme.

Performance of PRoWait on different mobility

models will also be explored in the future.

REFERENCES

Elizabeth M. Royer and Chai-Keong Toh, 1999, “A review

of current routing protocols for ad hoc wireless net-

works”, IEEE Personal Communication Magazine, vol.

6, no. 2, pp. 46-55.

C.-K. Toh, 2002, “Ad Hoc Mobile Wireless Networks: Pro-

tocols and Systems”, Prentice Hall PTR, Englewood

Cliffs, NJ.

L. Pelusi, A. Passarella, and M. Conti, 2006, “Opportunistic

networking: data forwarding in disconnected mobile ad

hoc networks”, IEEE Communications Magazine, vol.

44, Issue 11, pp. 134-141.

C.-M. Huang, K.-C. Lan, C.-Z, and Tsai, 2008, “A survey

of opportunistic networks”, in proceedings of the 22

Intl. Conference on Advanced Information Networking

and Applications- workshops, 2008 (AINAW 2008),

Okinawa, Japan, pp. 1672-1677.

S. K. Dhurandher, Deepak Kr. Sharma, I. Woungang, and

H.C. Chao, 2011, “Performance Evaluation of Various

Routing Protocols in Opportunistic Networks”, in Pro-

ceedings of IEEE GLOBECOM Workshop 2011, Hou-

ston, Texas, USA , pp. 1067-1071.

A. Keranen, 2008, “Opportunistic Network Environment

Simulator”, Special Assignment Report, Helsinki Uni-

versity of Technology, Dept. of Communications and

Networking.

A. Vahdat, and D. Becker, 2000, “Epidemic routing for par-

tially connected ad hoc networks”, Technical Report

CS-2000-06, Dept. of Computer Science, Duke Univer-

sity, Durham, NC.

T. Spyropoulos, K. Psounis and C. S. Raghavendra, 2005,

“Spray and Wait: An Efficient Routing Scheme for In-

termittently Connected Mobile Networks”, in proceed-

ings of SIGCOMM Workshop on Delay-Tolerant Net-

working, Philadelphia, USA, pp. 252-259.

A. Lindgren, A. Doria, and O. Schelen, 2003, “Probabilistic

routing in intermittently connected networks”, ACM

SIGMOBILE, Mobile Computing and Communica-

tions Review, vol. 7, Issue 3, pp. 19–20.

S. K. Dhurandher, Deepak Kr. Sharma, I. Woungang , and S.

Bhati, 2013, “HBPR: History Based Prediction for Rout-

ing in Infrastructure-less Opportunistic Networks”, in

Proceedings of 27

IEEE International Conference on

Advanced Information Networking and Applications

(AINA-2013), Barcelona, Spain, pp. 931-936.

S. K. Dhurandher, Deepak Kr. Sharma, I. Woungang, A.

Saini, 2014, “Efficient Routing based on Past Infor-

mation to Predict the Future Location for Message Pass-

ing in Infrastructure-less Opportunistic Networks”, in

the Special Issue Support Technology and Architecture

for Networked and Distributed Systems, Journal of Su-

percomputing, Springer, DOI: 10.1007/s11227-014-

1243-5, Online, June 28, 2014 (Impact Factor: 0.917).

S. K. Dhurandher, Deepak Kr. Sharma, I. Woungang, R.

Gupta, S. Garg, 2014, “GAER: Genetic Algorithm based

Energy-efficient Routing Protocol for Infrastructure-less

Opportunistic Networks”, Journal of Supercomputing,

Springer, DOI: 10.1007/s11227-014-1195-9, Online,

May 3, 2014 (Impact Factor: 0.917).

A. Keranen, J. Andott, 2007, “Opportunistic increasing re-

ality for DTN protocol simulations”, Special Technical

Report, Helsinki University of Technology, Networking

Laboratory.

WINSYS2015-InternationalConferenceonWirelessInformationNetworksandSystems