Experimental Synthesis of Routing Protocols and Synthetic Mobility

Modeling for MANET

Nisrine Ibadah

, Khalid Minaoui

, Mohammed Rziza

and Mohammed Oumsis

1,2

LRIT, Associated Unit to CNRST (URAC 29), Faculty of Sciences, Mohammed V University, Rabat, Morocco

Superior School of Technology, Mohammed V University, Sal

e, Morocco

Keywords:

MANET, Routing Protocols, Mobility Models, NS2, Boonmotion, Performance Analysis.

Abstract:

Many performance analyses are already done with a lot of ﬂaws. But, they do not look to all inﬂuenced con-

straints. In this study, we aim to summarize several parameters into 90 different scenarios with an average

of 1350 simulated ﬁles. That shows results of three performance metrics combined with ﬁve mobile ad hoc

routing protocols under three synthetic mobility models. All these parameters are applied to two dissimilar

simulation areas. Basing on one exhaustive analysis with all these details like this paper; leads to well under-

stand the accurate behaviors of routing protocols and mobility models used. By displaying the ability of every

routing protocol to deal with some topology changes, as well as to ensure network performances.

1 INTRODUCTION

For almost two decades, mobile communication has

become a major ﬁeld of research and scientiﬁc dis-

coveries. Mobile Ad hoc Network (MANET) has

achieved a huge improvement due to its ﬂexibility,

easier maintenance, the non-existence of centralized

control or ﬁxed and static infrastructure as well as

self-administration and self-conﬁguration abilities.

Several mobility models have been proposed to

overcome these situations with the aim of imitating

human beings’ real-life. Wireless communications

display many problems related to nodes density, traf-

ﬁc load, autonomous energy, and mobility. Routing

within this network suffers from frequent topology’s

updates and unconnected actives routes between mo-

bile nodes. The main challenge of MANETs rout-

ing is to develop a dynamic routing protocol expe-

ditiously able to ﬁnd a route between mobile nodes.

The choice of a mobility model (MM) can favorite

some designs over others. It must be efﬁciently

readapted to every change occurring in the network

topology(Srivastava et al., 2014). The performance

of mobile ad hoc networks can vary signiﬁcantly un-

der different mobility models. Sometimes, they eval-

uate routing protocols without taken into considera-

tion mobility models. They often analyze them using

one routing protocol. Whereas, Simulation time em-

ployed is too short. It mainly impacts performance

metrics of many mobility models. Or usually, sim-

ulation area used is small. It inﬂuences the number

of packets received. Their optimal implementation

requires a deep study of the routing protocols. Re-

searchers ﬁnd meaningful to explore mobility model

decisions and metrics in modeling their wireless com-

munication where mobile nodes move from a place to

another with no ﬁxed infrastructure.

Synthetic Mobility models(Umamaheswaran

et al., 2014) imitate the movement of real mobile

nodes that change speed, position and direction with

time. They can be done by making prevision, mobiles

move from one place to another at a given moment

under varied network restrictions. They represent

precisely motion characteristics of mobile nodes.

They are amongst key parameters that inﬂuence

performance features of the mobile network in

order to judge which protocol is useful in a special

scenario. Nodes’ mobility need to be analyzed to

explore dependency and topology requirements.

This paper will propose an intensive performance

analysis of some synthetic mobility model under a

mobile ad hoc network. In order to describe mobil-

ity issue of various wireless communication scenarios

that heavily impacts mobile routing protocols. The

entire document is divided into three principal sec-

tions. Firstly, we present brief related works where

are used in the simulation. Secondly, we present the

parameters of simulation; and also, we interpret the

simulation results. And ﬁnally, we discuss the con-

clusion.

168

Ibadah N., Minaoui K., Rziza M. and Oumsis M.

Experimental Synthesis of Routing Protocols and Synthetic Mobility Modeling for MANET.

DOI: 10.5220/0006203601680173

In Proceedings of the 6th International Conference on Sensor Networks (SENSORNETS 2017), pages 168-173

ISBN: 421065/17

2 RELATED WORKS

Many ways are proposed to classify synthetic mobil-

ity models(Batabyal and Bhaumik, 2015). Firstly the

’Entity mobility model’ where every node is indepen-

dent of each other. This class has been classiﬁed into

the following areas: random mobility models, mod-

els with temporal dependency, models with spatial

dependency and models with geographic restrictions.

For random mobility models, nodes travel freely and

without obstructions. Direction, speed, and destina-

tion are selected randomly and independently of prior

selection. That assesses these models to be generally

without a memory, e.g: Random Waypoint Mobility

Model (RWMM)(Han et al., 2016). However, mod-

els with geographic restriction, node’s movements are

not often random or have a temporal/spatial depen-

dency. But, it can be obstructed in a bounded area,

guided by paths or restricted into a building, e.g.,

Manhattan Grid Mobility Model (MGMM)(Martinez

et al., 2013). Secondly the ’correlated or group based

mobility model’, where the device node’s movement

is dependent on others. In this subclass, nodes move

by following a leader node in the group. That is to

say, each group is governed by one leader which can

be a pre-deﬁned or a logical node, e.g., Reference

Point Group Mobility Model (RPGMMM)(Dong and

Dargie, 2013). Thirdly the ’human or social based

mobility model’ where nodes are driven by social-

izing human behaviors, e.g., Self-Similar Least Ac-

tion Walk(Hiranandani et al., 2013). And fourthly,

vehicular mobility models emulate vehicle movement

with changing speed, moving in queues along high-

way/street and stopping at trafﬁc signals(Al-Sultan

et al., 2014). That follow the shortest trajectory from

a given source to a destination. However, vehicular

communication becomes an important portion of the

intelligent transport system.

3 SIMULATION PARAMETERS

AND RESULTS

3.1 Conﬁguration Parameters

This paper shows results of three performance met-

rics which are Packet Delivery Ratio (PDR), average

end-to-end delay and throughput under different sce-

narios. we combine ﬁve mobile ad hoc routing proto-

cols which two of them are proactive, two are reactive

and hybrid one. With three synthetic mobility models

which are: RWMM is a random entity synthetic MM,

MGMM is an entity synthetic MM with restriction ge-

ographic MM and RPGMM is group based MM. All

Table 1: Simulation parameters.

Parameters Values

Propagation model TwoRayGround model

Bandwidth 10 Mb/s

Number of nodes 50

Packet size CBR

Packet rate 512 bytes/s

Speed 10 m/s

Pause time (s) 0, 20, 40, 60, 80

Routing Protocols

DSDV, OLSR, AODV,

DSR, and ZRP

Mobility models

RWMM, MGMM, and

RPGMM

Performance metrics

PDR, Average e-e delay,

and Throughput

Area

220 * 220 , and 1020 *

1020

Simulation time 1000 s

Recursion 15 times

these parameters are applied under two simulation ar-

eas; small one with (220m*220m) and large one with

(1020m*1020m). So, our results will represent 90 dif-

ferent scenarios with an average of 1350 simulated

ﬁles. We combine all these details in order to well un-

derstand the accurate behaviors of routing protocols

and mobility models used. Simulation settings used

for the experiments are depicted in Table 1.

3.2 Results and Discussion

To evaluate routing protocols, a wide range of perfor-

mance metrics have been considered to catch charac-

teristics of different mobility models. Our results aim

to analyze their performance impacts on routing pro-

tocols over MANET. So, different metrics have been

used to compare and evaluate them against nodes’

mobility, as follows:

Firstly, we start with Packet Delivery Ratio (PDR) or

Fraction (PDF). It represents the ratio of data pack-

ets delivered to destinations, those generated by CBR

application sources. According to this metric, simu-

lation results are shown in Figure 1 and Figure 2.

Figure 1 is applied in the small area. From Fig-

ure 1 (a) and (c), the PDR of AODV and DSR present

best results in both RWMM and RPGMM in which

they reach approximately 100%. Due to their reac-

tive strategy, routes are sure which are searched on

demand. But, AODV represents the best routing pro-

tocol in MGMM of Figure 1 (b). However, in RWMM

Experimental Synthesis of Routing Protocols and Synthetic Mobility Modeling for MANET

169

and MGMM, ZRP gives the worst results in this

metric, by dint of zone network used by this proto-

col. DSDV and OLSR in RWMM and MGMM offer

acceptable outcomes, thanks to continuously update

their routing table. OLSR is the worst in RPGMM.

As a result of, OLSR is based on routing by cluster

heads. And, RPGMM has their own groups’ leader.

So, the same strategy applied for routing and mobility

respectively. The coordination between clusters and

leader nodes is tough in this case. In general, we no-

tice that AODV offers best results at the PDR for all

mobility used in the small area.

(a) RWMM

(b) MGMM

Figure 1: PDR of routing protocols under various mobil-

ity models - Small area. (a) Random Waypoint Mobility

Model, (b) Manhattan Grid Mobility Model, (c) Reference

Point Group Mobility Model.

Figure 2 is applied in the large area. From Figure

2 (a), (b) and (c), the DSR and ZRP offer the best PDR

percentage. Due to the hidden routing table of DSR

which often has an available route to the destination

even in a wide ﬁeld. And zone based protocol applied

by ZRP which it allows to be suitable to the large area.

Although, the proactive protocols OLSR and DSDV

are the worst in all mobility models. Proactive pro-

tocols generally offer bad results in large simulation

ﬁeld. We observe that ZRP is the best in the PDR

in this area. As a result of dividing spacious simula-

tion area in a small zone which will be easier to verify

transmitted packets.

(a) RWMM

(b) MGMM

Figure 2: PDR of routing protocols under various mobil-

ity models - Large area. (a) Random Waypoint Mobility

Model, (b) Manhattan Grid Mobility Model, (c) Reference

Point Group Mobility Model.

Secondly, we analyze the ’Average End-to-End

Delay’. It represents total time spends between ap-

plication source to destination one. The simulation

results are shown in Figure 3 and Figure 4.

Figure 3 is applied in a small area. From Fig-

ure 3 (a),(b) and (c), the Average end-to-end delay

of DSR and ZRP are the worst in this three mobility

models simulated. Due to their zone approach of ZRP

and useless routes saved by DSR. However, we notice

that in the small area, this metric is best with AODV,

OLSR, and DSDV. Thanks to their on demand or con-

tinuous proactive strategy adopted by these routing

protocols.

SENSORNETS 2017 - 6th International Conference on Sensor Networks

170

(a) RWMM

(b) MGMM

Figure 3: End-to-End Delay of routing protocols under var-

ious mobility models - Small area. (a) Random Waypoint

Mobility Model, (b) Manhattan Grid Mobility Model, (c)

Reference Point Group Mobility Model.

Figure 4 is applied in a wide area. From Figure

4 (a), (b) and (c) like the small one, the average end-

to-end delay of DSR and ZRP are the worst in these

three mobility models simulated of Figure 4 (a),(b)

and (c). Due to their zone approach of ZRP and use-

less routes saved by DSR. So, sometimes, they borrow

prolonged routes to reach the destination. However,

AODV has acceptable results. Thanks to the reactive

methodology which send to one neighbor without to-

tal knowledge of a correct path to the destination. We

notice that average end-to-end delay of proactive pro-

tocols OLSR and DSDV is not inﬂuenced by simula-

tion ﬁeld adopted. It offers best outcomes, thanks to

their continuous proactive strategy.

Thirdly, we assess the Throughput which is the

sum of data rates which are delivered to all mobile

nodes, indicating bits or packets received per second.

The simulation results are shown in Figure 5 and Fig-

ure 6.

(a) RWMM

(b) MGMM

Figure 4: End-to-End Delay of routing protocols under var-

ious mobility models - Large area. (a) Random Waypoint

Mobility Model, (b) Manhattan Grid Mobility Model, (c)

Reference Point Group Mobility Model.

Figure 5 is applied in a small area. From Figure 5 (a)

and (c), reactive protocols AODV and DSR show best

results in RWMM and RPGMM. These protocols are

suitable for small areas. But from Figure 5 (b), AODV

outperforms than others at MGMM due to the reliable

path used. However, ZRP is the worst in RWMM

and MGMM. And, it is admissible in RPGMM. Al-

though, DSDV and OLSR offer permissible outcomes

in RWMM and MGMM. But, OLSR is the worst in

RPGMM due to its cluster routing process. We con-

clude that AODV is the most suitable for all these mo-

bility models simulated in a small ﬁeld.

Figure 6 is applied in a large area. From Figure

(a), (b) and (c), we remark that ZRP, AODV, and DSR

gives best results on the throughput. Furthermore,

ZRP is the best according to this metric. But, OLSR

and DSDV are the worst at all models experimented.

We conclude that proactive protocols are bad. And,

ZRP is the best one in large areas.

Experimental Synthesis of Routing Protocols and Synthetic Mobility Modeling for MANET

171

(a) RWMM

(b) MGMM

Figure 5: Throughput of routing protocols under various

mobility models - Small area. (a) Random Waypoint Mo-

bility Model, (b) Manhattan Grid Mobility Model, (c) Ref-

erence Point Group Mobility Model.

After simulating 1350 ﬁles of 90 different sce-

narios. Our results will be summarized in Table 2.

When we are combined some routing protocols with

synthetic mobility models. We obtain best outcomes

which are displayed with green cells 1-2. And worst

results with red color 4-5.

We result from that the Packet delivery ratio and

Throughput in a small area. AODV achieve best out-

comes as a result of on-demand concept based on

route request RREQ and route reply RREP leads to

possesses exactly the correct path. But, the worst one

is represented by ZRP because it explore information

of Intra-zone Routing Protocol (IARP) and Inter-zone

Routing Protocol (IERP) which will be tedious to co-

ordinate between them in a small one.

For the large area, we acquire best results with DSR

due to the available path to a destination node, even if

in a wide area. And ZRP as a result of dividing spa-

cious simulation area in a small zone which will be

(a) RWMM

(b) MGMM

Figure 6: Throughput of routing protocols under various

mobility models - Large area. (a) Random Waypoint Mo-

bility Model, (b) Manhattan Grid Mobility Model, (c) Ref-

erence Point Group Mobility Model.

easier to verify transmitted packets.

However, for the average end-to-end delay. In the two

areas, we have best results with proactive protocols

DSDV and OLSR, due to their researches in advance

and continuous updates or routing tables. So, all the

time, they possess correct paths to a destination. But,

the worst are obtained with DSR as a result of hidden

table without any strategy to erase it, and ZRP due to

speed occupied to locate the destination in a speciﬁc

zone in simulation ﬁeld.

4 CONCLUSION

This paper aimed to summarize several performance

evaluation scenarios of MANET routing protocols

under different mobility models. Three mobility

models have been applied in order to study the im-

SENSORNETS 2017 - 6th International Conference on Sensor Networks

172

Table 2: Experimental synthesis results.

Performance Routing Mobility models

Metrics Protocols Small area LARGE area

RWMM MGMM RPGMM RWMM MGMM RPGMM

PDR AODV 2 1 1 3 3 3

DSR 1 4 2 1 2 2

DSDV 4 2 3 5 5 4

OLSR 3 3 5 4 4 5

ZRP 5 5 4 2 1 1

Avg AODV 3 3 3 3 3 3

e-to-e delay DSR 4 5 4 5 5 5

DSDV 2 1 2 1 2 1

OLSR 1 2 1 2 1 2

ZRP 5 4 5 4 4 4

Throughput AODV 2 1 1 3 3 3

DSR 1 4 2 2 2 2

DSDV 4 2 3 5 5 5

OLSR 3 3 5 4 4 4

ZRP 5 5 4 1 1 1

pact of changed metrics as average end-to-end de-

lay, throughput and the packet delivery ratio. We

conclude that AODV offers best results in the small

area. It is usually moderate or better for all ninety

divers’ scenarios. It represents an adaptable routing

protocol under varied mobility models for the small

and large area. Due to, its reactive routing approach

which leads it to own correct path according to pack-

ets transmitted. However, ZRP is the worst. But, it

is the best in the large one. And proactive protocols

are the worst in this ﬁeld. Three tracks in mobility

modeling are allowed which we achieve the ﬁrst one

in this paper. Basing on one itemized analysis with

all these details; leads to well understand the accu-

rate behaviors of routing protocols and mobility mod-

els used. Our future work will focus on modeling a

human trace mobility model applied in a real world

scenario. That will be interesting for mobile P2P ap-

plication and suitable to a crowded area.

ACKNOWLEDGEMENTS

This paper was supported by the project PPR

◦

13/2016 of Mohammed V University and LRIT

Laboratory, Rabat.

REFERENCES

Al-Sultan, S., Al-Doori, M. M., Al-Bayatti, A. H., and

Zedan, H. (2014). A comprehensive survey on vehicu-

lar ad hoc network. Journal of network and computer

applications, 37:380–392.

Batabyal, S. and Bhaumik, P. (2015). Mobility models,

traces and impact of mobility on opportunistic routing

algorithms: A survey. IEEE Communications Surveys

& Tutorials, 17(3):1679–1707.

Dong, Q. and Dargie, W. (2013). A survey on mobility and

mobility-aware mac protocols in wireless sensor net-

works. IEEE Communications Surveys & Tutorials,

15(1):88–100.

Han, G., Jiang, J., Zhang, C., Duong, T., Guizani, M., and

Karagiannidis, G. (2016). A survey on mobile anchor

node assisted localization in wireless sensor networks.

Hiranandani, D., Obraczka, K., and Garcia-Luna-Aceves, J.

(2013). Manet protocol simulations considered harm-

ful: the case for benchmarking. IEEE Wireless Com-

munications, 20(4):82–90.

Martinez, F. J., Fogue, M., Toh, C.-K., Cano, J.-C.,

Calafate, C. T., and Manzoni, P. (2013). Computer

simulations of vanets using realistic city topologies.

Wireless Personal Communications, 69(2):639–663.

Srivastava, A., Mishra, A., Upadhyay, B., and kumar Ya-

dav, A. (2014). Survey and overview of mobile ad-hoc

network routing protocols. In Advances in Engineer-

ing and Technology Research (ICAETR), 2014 Inter-

national Conference on, pages 1–6. IEEE.

Umamaheswaran, S., Kumar, N., and Biswas, P. (2014).

A survey of routing protocol design patterns for mo-

bile ad hoc networks. In Emerging Technology

Trends in Electronics, Communication and Network-

ing (ET2ECN), 2014 2nd International Conference

on, pages 1–6. IEEE.

Experimental Synthesis of Routing Protocols and Synthetic Mobility Modeling for MANET

173