Scalability of MANET Routing Protocols for Heterogeneous and Homogenous Networks

University of Nebraska - Lincoln University of Nebraska - Lincoln

DigitalCommons@University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Faculty Publications in Computer & Electronics

Engineering (to 2015)

Electrical & Computer Engineering, Department

12-2007

Scalability of MANET Routing Protocols for Heterogeneous and Scalability of MANET Routing Protocols for Heterogeneous and

Homogenous Networks Homogenous Networks

Huda Al Amri

University of Wollongong

Mehran Abolhasan

University of Wollongong

Tadeusz Wysocki

University of Nebraska-Lincoln

, wysocki@uow.edu.au

Follow this and additional works at: https://digitalcommons.unl.edu/computerelectronicfacpub

Part of the Computer Engineering Commons

Al Amri, Huda; Abolhasan, Mehran; and Wysocki, Tadeusz, "Scalability of MANET Routing Protocols for

Heterogeneous and Homogenous Networks" (2007).

Faculty Publications in Computer & Electronics

Engineering (to 2015)

. 41.

https://digitalcommons.unl.edu/computerelectronicfacpub/41

This Article is brought to you for free and open access by the Electrical & Computer Engineering, Department of at

DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Faculty Publications in

Computer & Electronics Engineering (to 2015) by an authorized administrator of DigitalCommons@University of

Nebraska - Lincoln.

Scalability of MANET Routing Protocols for

Heterogeneous and Homogenous Networks

Huda Al Amri, Mehran Abolhasan , Tadeusz Wysocki

Telecommunication and Information Research Institute, University of Wollongong, Wollongong, NSW 2522, Australia

Abstract- In Mobile Ad hoc Network (MANET), mobility, traffic

and node density are main network conditions that significantly

affect the performance of routing protocols. Much of the previous

researches in MANET routing have focused on developing

strategies, which suit one specific networking scenario. Therefore,

there is no existing protocol that can work well in all different

networking scenarios. This paper reviews characteristics of each

different classes of routing protocols. Moreover, most of current

routing protocols assume homogeneous networking conditions

where all nodes have the same capabilities and resources.

Although homogenous networks are easy to model and analysis,

they exhibits poor scalability compared with heterogeneous

networks that consist of different nodes with different resources.

This paper presents extensive studies simulations for DSR,

AODV, LAR1, FSR and WRP in homogenous and heterogeneous

networks. The results showed that these which all protocols

perform reasonably well in homogenous networking conditions,

their performance suffer significantly over heterogonous

networks.

I. INTRODUCTION

Mobile Ad hoc NETworks (MANET) are a group of wireless

mobile nodes that have no fixed infrastructure. Therefore each

node can act as router or an end-user node. Many routing

protocols have been proposed to mange the communication on

this kind of networking. Moreover, there are many issues that

must be considered in constructing any routing protocol such

as power consumption, reliable data delivery, and overheads

and delays.

Recent work on MANET routing protocols have focused on

achieving stability and reliability to reduce packet loss,

communication overheads, and to increase data delivery ratio.

Different approaches have been proposed to achieve those

goals. Some of those focused on improving physical layer to

provide reliable transmission, like diversity techniques, coding

and Single Path Parallel Relays (SPPR) strategies [1-3].

Cooperation between link layer and network layer was another

approach [3], where the state and the availability of the link on

link layer were analyzed before calculating the routes [3].

Others expanded the existing protocols like AODV, LAR, and

DSR by implementing the multipath strategy [4, 5]. However,

mobility of the nodes has not been the main focus of those

papers. We anticipate that several problems in MANETs arise

due to the mobility such as high data delay and low packet

delivery ratio. Hence, node mobility has to be considered in

order to achieve high stability and reliability. Different

strategies have been implemented in [6-8] to satisfy different

degrees of mobility. On the other hand, most existing routing

protocols have not been able to satisfy both scalability and

mobility. Many routing strategies have been proposed to

improve the performance of existing protocols or design new

ones to deal with mobility or node density. In [8], Adaptive

Cell Relay routing protocol (ACR) has been designed to deal

with different density degree of the nodes to achieve high

scalability. It uses two different routing strategies: the cell relay

(CR) routing for dense networks, the large cell (LC) routing for

sparse networks. It also monitors the node density changes to

determine which routing strategy to apply according to the

network density. A routing framework has been proposed in [6]

to work on different mobility classes that are low, normal and

fast. Mobility class was calculated using the proposed mobility

metric referred to as "Stability". Stability is based on

associativity that is defined as a time where the node can

communicate with other nodes, and according to the stability

value, a protocol is selected to route the packet. If a node is

classified to be slow then a proactive protocol, like DSDV, will

be used, if mobility class is normal then a reactive protocol like

AODV will be applies, and the introduced RUNNER protocol

will route the data if the mobility class is fast. In [9], two new

protocols have been proposed to work with high mobility

nodes in MANET. The idea behind these two protocols is that

there is a group of mobile nodes which move throughout the

entire network to receive and deliver data and control

messages. These nodes are called the support nodes. One of the

protocols is called Snake where the support nodes are

predefined and then a leader election is carried out. The leader

manages the movement of its group of support nodes in a form

of snake movement. While each support nodes in RUNNER

(second protocol) moves independently like a runner. However,

the idea of support protocols cannot be applied in homogenous

systems. Support nodes should have more capabilities and

resources.

Few comparisons between different existing protocols have

been published such as in [10, 11]. For example, in [10],

AODV protocol and RUNNNER protocol have been evaluated.

It has been found that AODV has higher data delivery ratio and

lower data delay in dense network and low mobility of nodes.

This is because AODV can reach destinations easily in such

network conditions. On the other hand, RUNNER performs

better in high mobility network where the support nodes are

faster in delivering data. In [11], DSDV, AODV, and DSR

have been compared in different scenarios of nodes mobility

and traffic loads. The simulation results showed that reactive

protocols (DSR and AODV) performed better than proactive

protocols when nodes were moving. In addition, DSR works

well with low traffic while AODV behaves better in higher

traffic. A probabilistic model has been proposed in [12] to

evaluate overhead of routing protocols of MANET. This model

depends on network topology and data traffic parameters to

estimate the number of control packets. In addition it can help

identifying a protocol for particular situation. This model was

tested by comparing it with existing simulations of AODV,

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

DSR, and OLSR. Reactive protocols again performed better

than proactive protocols when the mobility increases.

Most of current routing protocols assume homogeneous

network conditions where all nodes have the same capabilities

and resources. Although homogenous network are easy to

model and analysis, they exhibits poor scalability compared

with heterogeneous networks that consist of different nodes

with different resources. Heterogeneous MANET comprise of

mobile devices as Fig.1 that have different communications

capability such as radio range, battery life, data transmission

rate, etc. Moreover, in real world, some of MANET networks

are obviously heterogeneous like military battlefield networks

and rescue operations system. For instance, in a rescue

operations system as Fig.2, there are limited mobile devices

that are provided to individual rescuers, ambulances and police

vehicles, and helicobacter. Limited mobile devices have lowest

communication capabilities, while helicobacter is the most

powerful communication device which forms backbone of the

rescue team. Therefore, heterogeneity of nodes is another issue

that needs to be considered in constructing and developing

routing protocols for MANETs.

Figure. 1: Heterogeneous MANET.

Figure. 2: Heterogeneous MANET in rescue operations

System.

Recently, a few publications have introduced some strategies

to develop routing protocols to accommodate heterogeneous

MANETs. On-demand Utility-Based Routing Protocol

(OUBRP) strategy has been proposed in [13] to develop

reactive routing protocols to efficiently utilize the

heterogeneity of nodes. A utility-based route discovery

algorithm is used to choose the richest nodes with highest level

of resources during route discovery stage. The utility level of

resources is reduced, if the route was not found. OUBRP

reduces the number of re-broadcasting nodes. This strategy has

been implemented over AODV. It has been found that this

strategy improves routing discovery and reduces effect of route

failure. In [14], scalability issue of OLSR in heterogeneous

MANET has been studied. The study show that OLSR it does

not differentiate distinct nodes with different communication

capability and resources. This paper proposed a strategy to

optimize OLSR to be scalable over large heterogeneous

MANET. OLSR was improved by organizing nodes in

hierarchal structure. Hierarchal OLSR (HOLSR) has eliminated

overheads and reduces the size of routing table. With HOLSR,

the nodes are organized in logical level, where nodes with

lowest resources are in lower level. Each level has many

clusters, where the cluster head is a powerful node with highest

communication capability. HOLSR and flat OLSR have been

compared in terms of control overhead, computations

overhead, and end-to-end delay. HOLSR shows significant

improvement to the performance of OLSR. Also it performs

well in large heterogeneous MANET.

A new routing protocol was proposed in [15] to make use of

heterogeneity in MANET. The entire network area was divided

into cells with same size. The most powerful node in a cell acts

as a gateway, where most the routing load goes through. This

powerful node is called B-node. All B-nodes form the

backbone of the entire networking. B-nodes reduce number of

hops because they have high communication capability to

transmit data.

Heterogeneous MANETs have the potential of reducing the

amount of power used at user nodes. In [16] author state that

supply of power in heterogeneous wireless ad hoc networking

can affect the lifetime of network. They proposed a cross- layer

for Device Energy-Load Aware Relaying (DELAR) strategy to

utilize powerful nodes. This strategy suggested having a

schedule to use different transmission powers in different

periods. They also proposed "mini-routing" and Asymmetric

MAC (A-MAC) to support link level acknowledgements with

unidirectional links. The simulation of DELAR showed that

this strategy can reduce power consumption and increase

lifetime of network.

The common approach to dealing with heterogeneity of

nodes in previous papers [13-15] is to assign most of the

routing load for the powerful nodes, as they possess more

resources and communication capabilities. Consequently, this

approach eliminates number of hopes and can reduce delay.

However, this strategy may create critical problems if the

powerful nodes go off-line. For example, in a battlefield

network scenario, if a vehicle or a tank possessing more

powerful communication capabilities was destroyed, then the

communication with others including soldiers and vehicles

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

could be lost. Most publications have not considered this

situation. There should be alternative strategies to recover any

fault on powerful nodes that have been assigned as routers.

Up to now, several reviews have been published which

described the functionality and theoretical performance of

MANET routing protocols. For example, in [17, 18] routing

protocols for MANETs have been revised and classified

according their scalability. However, no study has attempted to

evaluate the performances of current routing protocols in

heterogeneous MANETs. In this paper, different classes of

MANET routing protocols are reviewed. A suitable class of

routing protocols is suggested to perform well in a particular

network conditions. Additionally, the performances of DSR,

AODV, LAR1, FSR and WRP are compared by simulating

them in homogenous and heterogeneous MANET.

In section two, a review of the key features of existing

routing protocols is summarized. The performance of these

protocols is analyzed in section three. Section four describes

our simulations of different protocols in homogenous and

heterogeneous MANET. Section five discusses our results.

Last section concludes this paper.

II. R

EVIEW OF MANET ROUTING PROTOCOLS

Routing protocols for MANETs have been classified

according to the strategies of discovering and maintaining

routes into three classes: proactive, reactive, and hybrid [17].

Of course, each routing protocol reacts differently to node

mobility and density. A routing protocol for MANETs is

usually evaluated in terms of performance metrics that are end-

to-end delay, overhead, throughput and data delivery ratio. This

section outlines the main features of each class. Also, a brief

summary about the protocols that have been used in

simulations is given.

Proactive Routing Protocols

Proactive routing protocols acquire routing information

periodically and store then in one or more routing tables. The

differences among the protocols in this class are routing

structure, number of tables, frequency of updates, use of hello

messages and the existence of a central node. Therefore, each

protocol reacts differently to topology changes. Flooding of

routing information is the mechanism that is often used to

discover and update routes. However, it is common that

proactive protocols generate more control traffic and overhead

than other protocol classes because of periodic updating which

increases as the number of nodes increases. Moreover, they

extensively use memory for storing those tables. Examples

proactive routing protocols are: Wireless Routing Protocol

(WRP) and Fisheye State Routing (FSR). With WRP [19], each

node maintains four routing tables. As the network increases,

this protocol consumes significant amount of memory.

Moreover, hello messages are used to ensure the connectivity

with neighbors. Consequently, this will consume more

bandwidth and power.

FSR [20] is another proactive protocol. This protocol updates

network information frequently for nodes that are within its

scope only. Therefore, it is more scalable than WRP. However,

FSR is not adaptable to high mobility.

Reactive Routing Protocols

Reactive routing protocols discover or maintain a route as

needed. This reduces overhead that is created by proactive

protocols [17, 21]. Flooding strategy is used to discover a

route. Reactive routing protocols can be classified into two

groups: source routing and hop by hop routing. In source

routing, data packet headers carry the path to destination.

Hence, intermediate nodes do not care about maintaining the

routing information. On the other hand, this kind of protocols

may experience high level of overhead as the number of

intermediate nodes increases. Also they have a higher chance

of a route failure. Packets in the second group of reactive

protocols have to carry only destination and next hop addresses

which means that nodes have to maintain and store routing

information for active routes. In general, reactive protocols

suffer from delay because of the route discovery process. Ad

hoc On-Demand Distance Vector (AODV) [17, 22] and

Dynamic Source Routing protocol (DSR), and Location-Aided

Routing (LAR) are well-known reactive routing protocols.

AODV is a hop-by-hop routing protocol, which introduces a

more dynamic strategy to discover and repair route when

compared to DSR. Destination sequence numbers are used to

avoid the problem of infinite loops. AODV maintains only

active routes to reduce overheads and control traffic. This

protocol is applicable for different levels of node density,

mobility and loads. It is suitable for scenarios with moderate

mobility and density networks.

DSR is a reactive source routing protocol [17, 23]. It

discovers routes on demands using route discovery and

maintenance strategy. Multiple routes are applied to achieve

load balancing and to increase robustness. DSR can operate

well with high mobility nodes because it can recover from

routes failure quickly. It can support up to one hundred node

which means it can work well over medium network density.

LAR [24] uses GPS information to detect the location of

nodes. This reduces overhead due to flooding. This protocol

has two strategies for route discovery. Firstly, it limits the

RREQ propagation for define area (i.e. Request Zone).

Secondly, stores the coordinates of destination node where the

route request packets travel toward the destination

coordinators. This reduces overheads. However, each node

must have GPS.

Hybrid Routing Protocols

Hybrid protocols exhibit both reactive and proactive

features. Proactive strategy is used to discover and maintain

routes to near by nodes, while routes to far away nodes are

discovered reactively. Consequently, overheads and delay that

are introduced by proactive protocols and reactive protocols,

respectively, are minimized. Hybrid protocols have been

known to be more scalable than others fewer nodes take part in

routing and topology discovery. In Zone Routing Protocol

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

(ZRP), the nodes are grouped into zones. Communications

between nodes depend on their locations in the zone. Another

example of hybrid protocol that can adapt to changes in node

density and mobility is Scalable Location Update Routing

Protocol (SLURP). It uses GPS information to mange node

location and eliminates global routing. Each node is associated

with a home region and sends its new location to its home

region as it moves .Hence, when a route is required, the source

node only have to query the home region of the destination.

This protocol is suitable for large networks where the number

of nodes and their mobility are high [17, 25].

Hierarchal and Geographic Routing Protocols

MANET routing protocols can be divided also according to

routing structures into flat routing protocols, hierarchal routing

protocols, and geographic position information assisted routing

protocols [18, 26]. Each protocol routes data proactively or

reactively or uses the combination of the two strategies. Flat

protocols can be tables driven (proactive) like DSDV and on

demand protocols (reactive) like DSR. Those protocols have

been described previously.

The idea of wireless hierarchal routing protocols is to group

mobile nodes to reduce the area of flooding. The nodes are

grouped in terms of clusters, trees or zones where there is a

leader that manages routing in its area. Each node has different

functionality depending on its location inside the group or

outside it. This strategy reduces the size of routing tables and

the routing information [18]. Example of wireless hierarchal

routing protocols is the ZRP. The advantages of those protocols

are in the reduction of overheads and improved scaling of large

networks compared to flat routing protocols. However, when

node mobility is high, hierarchal routing may introduce more

overhead due to cluster re-calculation. In addition, a cluster

head is a critical node and communication breaks if it goes

down.

Geographic position information assisted routing protocols

improve routing by using Global Position System (GPS)

receivers built into the nodes to get their location information

[18]. Those protocols route the data using Geographic

Addressing and Routing (GeoCast) where messages are sent to

all nodes in specific geographical area. GeoCast uses the

geographical information rather than logical addresses.

Geographical information about nodes eliminates propagation

of routing information. Hence, geographical protocols have

more efficiency in adapting to changes in node density

compared to other protocols. Examples of geographic routing

are DREAM and SLURP. However, mapping address to

location produces more overheads. In addition, using GPS

consumes the power of a mobile node.

III. C

OMPARISON OF DIFFERENT CLASSES OF ROUTING

PROTOCOLS FOR

MANETS

In this section, comparison of proactive, reactive and hybrid

protocols is outlined by combining their published theoretical

performance [27] [12]. That comparison is further verified

through the published simulation results [6, 11, 27-30]. Based

on that comparison, a suitable class of routing protocols is

suggested to perform well in a particular network conditions.

A. Theoretical and model based analysis

Proactive protocols are the oldest protocols that have been

derived from wired network routing protocols to work in the

wireless environment. Therefore, they possess many features of

wired routing protocols like routing tables that are used to keep

the routing information, which are periodically updated even if

not needed. As the node moves, there is a flooding of packets

containing the topology changes causing high overheads.

Hence, in general, proactive protocols produce more overheads

resulting in a lower throughput in case of high mobility as

illustrated in theoretical and model based analysis below.

In order to compare the protocols, the following set of

parameters is usually defined:

N=number of nodes.

L=average path length (in hops).

R=average number of active routes per node.

µ=average number of link breakage per second (reflect

mobility degree).

α=route activity, which gives how the frequently the node is

changing its destination.

ρ=route concentration factor that monitors the traffic hot-

spots in MANET.

Proactive, reactive and hybrid protocols have been evaluated

theoretically in [27]. It has been found that asymptotic

overhead for proactive is O(N

1.5

) due to the process of

maintaining and forwarding tables to keep periodic updates. In

reactive protocols, route requests and reply messages create

overhead of cost O(N

), while in hybrid protocols this is

O(N

1.66

). The number of packets that are produced by proactive

protocols per second is µ*L*N

while for reactive protocols is

(α+ρ*R*µ)*L*N

. Reactive is found to be better than proactive

if µ*L*N

> (α+ρ*R*µ)*L*N

. It has been concluded in [25]

that proactive protocols can be used mostly in static or quasi-

static networks, reactive protocols are preferred in more

dynamic networking, while hybrid protocols are more efficient

in adapting to changes in network conditions.

Analytical model that compared control overhead with

mobility and data traffic for proactive and reactive protocols

for MANETs has been also presented in [12]. It has been found

that number of packets produced by optimized reactive

protocols in MANET is o

µaLN

and o

µANpN

for optimized

proactive protocols, where

= route request optimization factor.

=active next hops ratio.

a= number of active routes per node (activity).

= broadcast optimization factor.

As a result of comparing those two approaches with existing

simulations, it has been observed that OLSR is more scalable

than DSR. Moreover, rough high mobility asymptotic for both

classes have been compared. It has been found that reactive

protocols are better than proactive in high mobility if reactive

protocols use routes that do not share links.

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

Hierarchal routing protocols, geographic position

information assisted routing protocols, and hybrid routing

protocols are more adaptable to various node destination than

flat protocols [17, 18]. In [17], hierarchal routing protocols

have been found to be more scalable than flat protocols

because they limit the propagation area by structuring the

network nodes. However, overheads are increasing with those

routing schemes due to location management. Therefore,

hierarchal protocols are suitable in scenario like high density

but low mobility. Geographic routing protocols also perform

well in high density because of the simplicity of location

management localized route discovery.

B. Simulation Observations

MANET routing protocols are commonly evaluated

according to performance metrics such as: delay, delivery ratio,

and overheads. Delay is the delay of data processing and

queuing in intermediate nodes. Delay increases usually as

mobility increases in all different classes of routing protocols.

The delivery ratio is the ratio of the number of received packets

at the destination to the number of packets that are sent by the

source node. This ratio usually decreases as mobility increases.

The last metric is the overhead consuming the network

bandwidth, which is often high as nodes increase their speed.

Adaptable protocol to particular scenario of density and

mobility has lower delay and overhead and higher delivery

ratio. Several simulations have been carried out to compare

different protocols from different classes in different scenarios

of nodes mobility and density [6, 11, 27-30].

The results of these simulations indicated that proactive

protocols have higher overhead than reactive and hybrid

protocols in terms of mobility and density while they have

smaller delay than reactive ones. On the other hand, reactive

protocols have lower delay than hybrid protocols. Although it

is noticed that as the density increases and the mobility

decreases, the delivery ratio increases. Proactive protocols have

better delivery ratio but hybrid protocols have the best delivery

ratio. Hence, they perform better in high density networks.

In [30], several simulations of four protocols have been

carried out using GloMoSim simulator. These protocols were

distance vector (DV), DSR and AODV as reactive protocols,

and WRP as proactive protocol. The simulations have been run

under different network conditions like different mobility

degrees and different nodes density. It has been found that

DSR has highest delay, while WRP has the lowest overhead as

mobility increases.

To conclude what we have outlined theoretically and from

existing simulations, proactive protocols class perform well in

network with low mobility nodes. However, this class can

adapt different node density, because they include hierarchal

and geographical routing protocols. Moreover, hierarchal,

geographic and, and hybrid routing protocols, have been more

flexible with high density networks. Therefore, they can

operate with medium and high density. In medium density and

mobility, reactive protocols can work well.

IV. S

IMULATION MODEL

In this section we present simulations that have been carried

out to compare the performances of different protocols from

different classes in heterogeneous and homogenous MANET.

In homogenous MANETs, all nodes have same capabilities and

resources while with heterogeneous MANET different nodes

have different resources like transmission range and power

saving.

We preformed the simulations using the GloMoSim [31]

package. Each simulation run for 900s with different values of

seeds. There was a 50-node network on a 1500x300 grid, a

100-node network on a 2200x600 grid, and a 200-node

network on a 3000x1000 grid. Random way point was used as

mobility model with eight different values of pause times that

were 0, 50, 100, 200, 300, 500, 700 and 900. Speeds of the

nodes were varied from 1 to 20 m/s. Different traffic loads

flows have been created between random pairs of nodes. There

were 10 flows from source to destination over 50-node, 110

flows over 100-node and 210 flows over 200-node. Constant

Bite Rate (CBR) was used to generate data traffic of rate 4

packets per second. Each packet was 512 bytes and transmitted

at 250 ms intervals.

IEEE 802.11 was used as MAC protocol with constant

transmission bandwidth of 2Mbps. The transmission power

was 15dbm for all nodes in homogenous network. In

heterogeneous MANET, nodes have different transmission

powers (1-20 dbm) and receiver powers (-81.0 - -110.0 dbm).

The simulations run five different protocols that were DSR,

AODV, LAR1, FSR and WRP. Packet delivery ratio (PDR),

End-to-End Delay and packet loose percentage, control

overhead and hop counts were used as performance metrics of

each protocol.

V. R

ESULTS

In this section, we present the results of simulating DSR,

AODV, FSR, LAR1 and WRP with different number of nodes

within heterogeneous and homogenous networks. DSR, AODV

and LAR1 were simulated with 50, 100 and 200 nodes while

FSR has simulated with 50 and 100 nodes and WRP with 50

only. This is because FSR and WRP are not scalable to large

number of nodes.

Fig 3 (a-e) shows End-to-End delay of each protocol. This

graph shows that all protocols behave differently with

heterogeneous nodes. The delay with 50 and 100 homogenous

nodes are nearly constant. They all have higher delay in

heterogeneous MANET comparing with homogenous MANET

and with same number of nodes. For example, AODV has

delay less than 0.2 with 50 homogenous nodes while it is more

than 0.3 with 50 heterogeneous nodes. Also, the delay is

extremely high with 200 heterogeneous nodes for all protocols.

DSR behaves better in heterogonous MANT than other

protocols do where delay is very low except for 200

heterogeneous nodes as shown in Fig3(b). FSR has high delays

with 100 nodes for both homogenous and heterogeneous

networks as in Fig.3(c). Fig.3 (d) illustrates the delay for

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

LAR1, all delays of different number of nodes in

heterogeneous MANET are the highest where it varies from 6

to 10 comparing with other protocols. All these reflect that

current routing protocols are not adaptable for heterogeneous

nodes.

Fig 4 (a-e) illustrate the packet delivery ratio (PDR) of

different protocols with homogenous and heterogeneous

networks. PDR of all protocols is nearly 1 in homogenous

networking with different number of nodes. It decreases with

heterogeneous networking. The difference between the PDR in

homogenous and in heterogeneous networks with same number

of nodes is higher with proactive protocols like FSR and WRP.

This difference is 20% for reactive protocols while it is nearly

60% with proactive protocols. This shows that those protocols

do not make use of the different resources that different nodes

have it.

(a): End-to-End Delay of AODV for 50, 100, and 200 homogenous and

heterogeneous nodes.

(b): End-to-End Delay of DSR for 50, 100, and 200 homogenous and

heterogeneous nodes.

(c): End-to-End Delay of FSR for 50 and 100 homogenous and heterogeneous

nodes.

(d): End-to-End Delay of LAR1 for 50, 100, and 200 homogenous and

heterogeneous nodes.

(e): End-to-End Delay of WRP for 50 homogenous and heterogeneous nodes.

Figure.3. (a-e): End-to-End Delay of AODV, DSR, FSR, LAR1, and WRP for

both homogenous and heterogeneous networks.

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

Packet loss percentage is illustrated in Fig 5(a-e). In

homogenous network, the rate of packet loss is very low

compared to heterogeneous networking. Packet loss percentage

in heterogeneous networking with reactive protocols is

between 20 and 25 while it ranges from 60 to 70 for proactive

protocols.

Overheads are illustrated in Fig 6(a-e). Overhead is higher

too with heterogeneous networking. 200 nodes with

heterogeneous networking has very high overhead with

AODV, DSR and LAR1. This illustrates the scalability issue.

Proactive protocols as expected have the highest overhead in

both homogenous and heterogeneous networking. This is

because of periodical updating of routing information.

(a): Packet Delivery Ratio of AODV for 50, 100, and 200 homogenous and

heterogeneous nodes.

(b): Packet Delivery Ratio of DSR for 50, 100, and 200 homogenous and

heterogeneous nodes.

(c): Packet Delivery Ratio of FSR for 50, and 100 homogenous and

heterogeneous nodes.

(d): Packet Delivery Ratio of LAR1 for 50, 100, and 200 homogenous and

heterogeneous nodes.

(e): Packet Delivery Ratio of WRP for 50 homogenous and heterogeneous

nodes.

Figure.4. (a-e): Packet Delivery Ratio of AODV, DSR, FSR, LAR1, and WRP

for both homogenous and heterogeneous networks.

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

Generally, most protocols behave inefficiently and

unexpectedly in heterogeneous networks. One of the problems

that cause misbehaving is unidirectional link. Some protocols

support only bidirectional link between two similar nodes.

Unidirectional link problem is shown in Fig 7, where node B

has higher transmission range than node A. Therefore

B includes A in its transmission range while A does not

include B. Consequently, the link between B and A is

unidirectional from B to A only. However, AODV assumes all

links between two nodes are bidirectional which gives incorrect

routing information. Therefore, this incorrect information

creates large delay and packet loss in heterogeneous

networking.

However, in heterogeneous networking, there are nodes

which have high transmission range to connect to large number

of nodes. Therefore, the number of neighbor nodes increases.

Hence, as network size increases, powerful nodes will consume

more memory and bandwidth in storing neighbor tables and

updating routing information. Therefore, proactive protocols

might experience higher percentage of packet losing and lower

PDR.

(a): Packet Loss Percentage of AODV for 50, 100, and 200 homogenous and

heterogeneous nodes.

(b): Packet Loss Percentage of DSR for 50, 100, and 200 homogenous and

heterogeneous nodes.

(c): Packet Loss Percentage of FSR for 50, and 100 homogenous and

heterogeneous nodes.

(d): Packet Loss Percentage of LAR1 for 50, 100, and 200 homogenous and

heterogeneous nodes.

(e): Packet Loss Percentage of WRP for 50 homogenous and heterogeneous

nodes.

Figure.5. (a-e): Packet Loss Percentage of AODV, DSR, FSR, LAR1, and

WRP for both homogenous and heterogeneous networks.

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

(a): Control Overhead of AODV for 50, 100, and 200 homogenous and

heterogeneous nodes.

(b): Control Overhead of DSR for 50, 100, and 200 homogenous and

heterogeneous nodes.

(c): Control Overhead of FSR for 50, and 100 homogenous and heterogeneous

nodes.

Figure.6. (a-e): Control Overhead AODV, DSR, FSR, LAR1, and WRP for

both homogenous and heterogeneous networks.

(d): Control Overhead of LAR1 for 50, 100, and 200 homogenous and

heterogeneous nodes.

(e): Control Overhead of WRP for 50 homogenous and heterogeneous nodes.

Figure.7: Unidirectional Link Problem in Heterogeneous MANET.

VI. CONCLUSION

In Mobile Ad hoc Network (MANET), mobility, traffic and

node density are main network conditions that significantly

affect the performance of the network. This issue has been

reviewed in this paper. In addition, most of current routing

protocols assume homogeneous network conditions where all

nodes have the same capabilities and resources. Although

homogenous networks are easy to model and analysis, they

exhibits poor scalability compared to heterogeneous networks,

which consist of different nodes with different resources. In

this paper, different simulations have been carried out to

compare the performance of different routing protocols in

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)

homogenous and heterogeneous networks. All simulated

protocols misbehave in heterogeneous networks. They also

suffer from high delays and achieve very low PDR. Current

MANET routing protocols have unidirectional link problem

and do not scale well if node density is increasing. This shows

that the current routing protocols for MANET are inadaptable

for heterogeneous networking. More works needs to be carried

out to investigate the problems that rise with routing protocols

in heterogeneous networking. In addition, most publications of

routing issues in heterogeneous MANET proposed to assign

most routing loads to powerful nodes. However, those nodes

may create critical problems if they go off-line. Hence, there

should be alternative strategies to recover any fault on powerful

nodes.

EFERENCES

1. Zhenzhen, Y. and H. Yingbo. Networking by parallel relays: diversity,

lifetime and routing overhead. 2004.

2. Zhenzhen, Y. and H. Yingbo. Stability of wireless relays in mobile ad

hoc networks. 2005.

3. Erik, W., et al., Improving routing performance in wireless ad hoc

networks using cross-layer interactions. Ad Hoc Netw., 2007. 5(5): p.

579-599.

4. Khan, A.Y., S. Rashid, and A. Iqbal. Mobility vs. predictive MPR

selection for mobile ad hoc networks using OLSR. 2005.

5. Valera, A., W.K.G. Seah, and S.V. Rao. Cooperative packet caching

and shortest multipath routing in mobile ad hoc networks. 2003.

6. Athanasios, B., et al., A mobility sensitive approach for efficient

routing in ad hoc mobile networks, in Proceedings of the 9th ACM

international symposium on Modeling analysis and simulation of

wireless and mobile systems. 2006, ACM Press: Terromolinos, Spain.

7. Mueller, S. and D. Ghosal. Analysis of a distributed algorithm to

determine multiple routes with path diversity in ad hoc networks.

2005.

8. Xiaojiang, D. and W. Dapeng, Adaptive cell relay routing protocol for

mobile ad hoc networks. Vehicular Technology, IEEE Transactions

on, 2006. 55(1): p. 278-285.

9. Ioannis, C., et al., An Experimental Study of Basic Communication

Protocols in Ad-hoc Mobile Networks, in Proceedings of the 5th

International Workshop on Algorithm Engineering. 2001, Springer-

Verlag.

10. Ioannis, C., K. Elena, and E.N. Sotiris, On the effect of user mobility

and density on the performance of protocols for ad-hoc mobile

networks: Research Articles. Wirel. Commun. Mob. Comput., 2004.

4(6): p. 609-621.

11. Per, J., et al., Scenario-based performance analysis of routing

protocols for mobile ad-hoc networks, in Proceedings of the 5th

annual ACM/IEEE international conference on Mobile computing and

networking. 1999, ACM Press: Seattle, Washington, United States.

12. Laurent, V., J. Philippe, and C. Thomas Heide, Analyzing control

traffic overhead versus mobility and data traffic activity in mobile Ad-

Hoc network protocols. Wirel. Netw., 2004. 10(4): p. 447-455.

13. Abolhasan, M., J. Lipman, and J. Chicharo. A routing strategy for

heterogeneous mobile ad hoc networks. in Emerging Technologies:

Frontiers of Mobile and Wireless Communication, 2004. Proceedings

of the IEEE 6th Circuits and Systems Symposium on 2004.

14. Villasenor-Gonzalez, L., G. Ying, and L. Lament, HOLSR: a

hierarchical proactive routing mechanism for mobile ad hoc networks.

Communications Magazine, IEEE, 2005. 43(7): p. 118-125.

15. Du, X. and D. Wu. Efficient multi-class routing protocol for

heterogeneous mobile ad hoc networks. 2005.

16. Wei, L., et al. DELAR: device/energy/load aware relaying in

heterogeneous wireless ad hoc networks. 2004.

17. Abolhasan, M., T. Wysocki, and E. Dutkiewciz, A

review of routing

protocols for mobile ad hoc networks. Elsevier journal of Ad hoc

Networks, 2004. l2(1): p. 1-22.

18. Xiaoyan, H., X. Kaixin, and M. Gerla, Scalable routing protocols for

mobile ad hoc networks. Network, IEEE, 2002. 16(4): p. 11-21.

19. Shree, M. and J.J. Garcia-Luna-Aceves, A routing protocol for packet

radio networks, in Proceedings of the 1st annual international

conference on Mobile computing and networking. 1995, ACM Press:

Berkeley, California, United States.

20. Gerla, M., Fisheye state routing protocol (FSR) for ad hoc network.

2002, IETF Internet Draft

21. Royer, E.M. and T. Chai-Keong, A review of current routing protocols

for ad hoc mobile wireless networks. Personal Communications, IEEE

[see also IEEE Wireless Communications], 1999. 6(2): p. 46-55.

22. Perkins, C., E. Belding-Royer, and S. Das, Ad hoc On-Demand

Distance Vector (AODV) Routing. 2003: RFC Editor.

23. Johnson, D., Y. Hu, and D. Maltz, The Dynamic Source Routing

Protocol (DSR) for Mobile Ad Hoc Networks for IPv4. February 2007:

RFC Editor

24. Young-Bae, K. and H.V. Nitin, Location-aided routing (LAR) in

mobile ad hoc networks. Wirel. Netw., 2000. 6(4): p. 307-321.

25. Seung-Chul, M.W. and S. Suresh, Scalable routing protocol for ad

hoc networks. Wirel. Netw., 2001. 7(5): p. 513-529.

26. Rajmohan, R., Topology control and routing in ad hoc networks: a

survey. SIGACT News, 2002. 33(2): p. 60-73.

27. Sholander, P., et al. Experimental comparison of hybrid and proactive

MANET routing protocols. 2002.

28. Samir, R.D., et al., Simulation-based performance evaluation of

routing protocols for mobile ad hoc networks. Mob. Netw. Appl.,

2000. 5(3): p. 179-189.

29. Ioannis, C., K. Panagiotis, and Z. Christos, Routing protocols for

efficient communication in wireless ad-hoc networks, in Proceedings

of the 3rd ACM international workshop on Performance evaluation of

wireless ad hoc, sensor and ubiquitous networks. 2006, ACM Press:

Terromolinos, Spain.

30. Ashwini, K.P. and F. Hiroshi, Study of MANET routing protocols by

GloMoSim simulator. Int. J. Netw. Manag., 2005. 15(6): p. 393-410.

31. Glomosim scalable simulation environment for wireless and wired

network systems. 2003 [cited; Available from:

http://pcl.c.ucla.edu/sprojects/glomosim/

Al Amri, Abolhasan & Wysocki in Proceedings og the International Conference on Signal Processing and Communication Systems (2007)