A new dynamic counter-based broadcasting scheme for

Full text

A new dynamic counter-based broadcasting scheme for
Mobile Ad hoc Networks
Muneer Bani Yassein
a
, Sanabel Fathi Nimer
a
, Ahmed Y. Al-Dubai
b,
⇑
a
Department of Computer Science, Jordan University of Science and Technology, P.O. Box 3030, Irbid, 22110, Jordan
b
School of Computing, Edinburgh Napier University, Edinburgh EH10 5DT, UK
article info
Article history:
Received 16 March 2010
Received in revised form 24 August 2010
Accepted 25 August 2010
Available online 31 August 2010
Keywords:
MANETs
Broadcast
Flooding
Counter-based
Broadcast storm problem
Reachability
Delivery ratio
abstract
Broadcasting is an essential operation in Mobile Ad hoc Networks (MANETs) to transmit a
message (data packet) from the sender to the rest of the network nodes. Although flooding
is the simplest mechanism for broadcasting, where each node retransmits every uniquely
received message exactly once, it is usually costly and results in serious redundancy, con-
tention and collisions in the network. These problems are widely referred to as the broad-
cast storm problem. In the light of this, this study introduces a new counter-based
broadcasting scheme to achieve efficient broadcasting in MANETs. This is achieved by
using a counter-based scheme with a dynamic threshold to increase the successful delivery
rate of packets and enhance the throughput of the network. Extensive simulation experi-
ments have been conducted. Our results show that the new scheme outperforms the well
known exiting schemes, namely the two counter-based broadcasting scheme and blind
flooding.
Crown Copyright Ó2010 Published by Elsevier B.V. All rights reserved.
1. Introduction
A Mobile Ad hoc Network (MANET) is an independent system consisting of a set of wireless mobile nodes, which com-
municate with each other without the existence of infrastructure. MANETs has several characteristics. First, the node in
the MANETs is self-organizing and self-administrating without deploying any infrastructure. Second, MANET mobile nodes
communicate with each other using multi-hop wireless links. Third, MANET topology changes could occur randomly, rapidly
and frequently, so the topology is dynamic [1]. There are number of characteristics in MANET such as mobility services, no
infrastructure and battery-powered properties make it used in a number of applications for MANET. MANET is widely used in
military, emergency operations, battle-fields, disaster recovery, group communication, civil and business operations. MAN-
ETs can be very useful in setting up an infrastructure-less network used to make a reliable and fast communication among
soldiers in the battle-fields to recover any failure in the network.
In MANETs, broadcasting plays a fundamental role as a means of broadcasting a message from a source node to all other
nodes in the network [2]. Broadcasting is a fundamental operation in several applications such as discovering neighbors, pag-
ing, addressing, communication in battlefield, home networking, temporary local area networks, disaster recovery opera-
tions and group communication [3–5]. Moreover, there are many routing protocols that use broadcasting for route
discovery [4] such as Ad hoc On-Demand Distance Vector Routing (AODV) [6], Dynamic Source Routing (DSR) [6], Zone Rout-
ing Protocol (ZRP) [7], and Location Aided Routing (LAR) [8]. In this paper, we will add a new approach to the AODV protocol
with counter-based scheme to transmit the packets to all the nodes with optimal packets delivery ratio.
1569-190X/$ - see front matter Crown Copyright Ó2010 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.simpat.2010.08.011
⇑
Corresponding author. Tel.: +44 131 4552796; fax: +44 131 4552727.
E-mail addresses: masadeh@just.edu.jo (M.B. Yassein), cs_sanabelfn@yahoo.com (Sanabel Fathi Nimer), a.al-dubai@napier.ac.uk (A.Y. Al-Dubai).
Simulation Modelling Practice and Theory 19 (2011) 553–563
Contents lists available at ScienceDirect
Simulation Modelling Practice and Theory
journal homepage: www.elsevier.com/locate/simpat

AODV is an on-demand protocol used to provide the route discovery and maintenance in a wide variety of network topol-
ogies and environments and to achieve improved performance, robustness and better scalability. Route Discovery works like
this: When a source node needs to communicate with a particular destination, it checks its routing table for the existence of
a path towards this destination in. In case a route is found, then it transmits the data to this destination, otherwise, a route
discovery procedure is evolved. The source node creates and broadcasts a Route Request (RREQ) packet to reach to the des-
tination itself or an intermediate node with a ‘fresh enough’ route to the destination as a valid route entry for the destination
whose sequence number is similar to that contained in the RREQ. Each node receiving the request sends a reverse route by
unicasting a Route Reply (RREP) back to the source. Route Maintenance is a mechanism used to repair routes when they are
invalidated or have broken links, so this error propagated to neighbors that have used this node as their next hop. It then
creates a Route Error (RERR) message and propagated to all nodes until it reach to the source node. Once the source receives
the RERR, it can re-initiate a route discovery if it still requires the route.
The main feature of AODV is its ability to use a destination sequence number for each route entry created by the desti-
nation for any route information sending to requesting nodes with loop freedom and this requesting node always selects the
node with the greatest sequence number. This protocol works on wired and wireless media. In AODV, the neighboring nodes
can detect each other’s broadcasts by using symmetric links between neighboring nodes. It does not attempt to follow paths
between nodes when one of these nodes cannot hear the other one [9].
These protocols are based on simplistic form of broadcasting called flooding; where every node in the network retransmits
every unique received packet exactly once but may lead to a serious problem, often known as the broadcast storm problem
[5,10]. So to solve this problem, the researchers present two categories. The first category is known as probabilistic broadcast
schemes and includes probability-based, counter-based, location-based, distance-based and hybrid-based schemes.
In probability-based scheme, the node rebroadcasts a message according to fixed and predetermined probability
around 0.65. In counter-based schemes, messages are rebroadcasted only when the number of copies of the message re-
ceived at a node is less than a threshold value. In the location-based scheme, messages are rebroadcasted only when the
additional coverage concept [11] determines the location of the mobile nodes to broadcast. In distance-based scheme mes-
sages are rebroadcasted according to the decision made between the relative distance of mobile node and the previous
sender.
In cluster-based scheme, the network is divided into number of clusters; each cluster has a single cluster head and several
gateways. Each cluster head, in turn, acts as a source for rebroadcast within its own cluster and the gateways can commu-
nicate with external clusters and are responsible for transmitting the broadcast message externally. Hybrid schemes [12,13]
combine between the advantages of probabilistic and counter-based schemes to achieve the performance improvement. The
second category is known as a deterministic broadcast scheme and includes multipoint relaying [14], node-forwarding [15],
neighbor elimination [16], and clustering [17]. Deterministic schemes use network topological information to build a net-
work including all the nodes in the network, so every node needs to exchange its information. Probabilistic schemes do
not use any information from network but, rather, start of building a network with each broadcast domain. Consequently,
these schemes have smaller overhead than deterministic ones.
The rest of this paper is organized as follows. Section 2introduces the related work of the counter-based rebroadcast. Sec-
tion 3presents the new algorithm proposed of counter-based scheme. Section 4evaluates the potential characteristics of our
scheme. Finally, Section 5concludes this study.
2. Related work
Flooding has been one of the earliest broadcast mechanisms, where every node in the network retransmits a message to
its neighbors when receiving it for the first time, but this mechanism cause a very important problem known in the MANETs
environment as the broadcast storm problem [3,5,10,18]. The probabilistic approach has been proposed in [5,11,18,19] as a
mechanism to reduce redundant rebroadcast messages [20]. Probabilistic approach works as follows: when receiving a pack-
et, each node forwards the packet with probability p. Ni et al. [10] have proposed a probability-based scheme to reduced the
redundant rebroadcast packets like flooding and counter-based schemes. Every node in flooding is rebroadcast with a fixed
probability P. On the other hand, counter-based scheme is proposed with additional coverage of each rebroadcast when
receiving nredundant messages of the same packet.
Zhang and Agrawal in [21] proposed a Dynamic probabilistic broadcast scheme as a combination of the probabilistic and
counter-based approaches. The scheme is implemented using AODV protocol. Cartigny and Simplot in [11] proposed the
Probabilistic scheme as a combination of the advantages of probability-based and distance-based schemes.
Another approach has been developed in order to deal with broadcast message called Counter-based scheme which works
as follows: when receiving a packet, the node initiates a counter and a timer. The counter is increased by one for each
received redundant packet. When the timer terminates, if the counter is larger than a threshold value, the node will not re-
broadcast the packet; otherwise, the node will broadcast it. On the other hand, when receiving a packet in the distance-based
scheme using the timer to know the locations of the senders of each received packet. Before the timer terminates, the node
checks the location of these senders. If any sender is closer than a threshold distance value, the node will not rebroadcast the
packet. Otherwise, the node rebroadcasts it. After that, another type of schemes was proposed to reduce the redundant
rebroadcast as Location-based scheme and it works as follows: when receiving a packet, the node initiates a timer and a
554 M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563

coverage area of packet. When the timer terminates, if the coverage area of arrived packet is larger than a threshold value,
then the node will not rebroadcast the packet. Otherwise, the node will broadcast it.
Counter-based broadcasting was proposed in [5,9,10] as a mechanism to reduce redundant rebroadcast messages and
solve the problems appeared in flooding. Counter-based scheme is based on the number of duplicate broadcast messages
received. The counter-based broadcasting scheme works as follows: upon receiving a message for the first time a counter
cis set to count the number of the duplicate messages received. A random assessment delay (RAD) timer is set. The RAD
is a timer randomly chosen between 0 and T
max
seconds, where T
max
is the maximum time delay used with randomized
scheduling to prevent collisions [22], when the RAD timer terminated the counter cis compared with fixed threshold value
Cto know if the node rebroadcast or not.
An efficient counter-based scheme has been proposed in [13,19] which combines the advantages of probability-based and
counter-based algorithms using a rebroadcast probability value of around 0.65, each node rebroadcasts that packet accord-
ing to this probability with counter threshold to enhance saved-rebroadcast, end-to-end delay and reachability. However, in
[12], they used rebroadcast probability value around 0.5 and achieve better performance than other schemes.
Ni et al. continue in [18] to describe an adaptive counter-based scheme in which each node dynamically captures its
threshold value Cbased on its number of neighbors by changing the fixed threshold Cinto a function C(n). In this approach
each node needs to estimate the current value of nwhere nis the number of neighbors of a node [3,5,10]. There exists a
method used to gather information about neighbors of any node by exchanging ‘Hello’ packets between neighbors to con-
struct a neighbor list at the nodes. A high number of neighbors implies that the node in a dense area but the low number of
neighbors implies that the node in a sparse area.
Williams and Camp [19] have partitioned the broadcast protocols into flooding, probability-based, counter-based, dis-
tance-based, location-based, clustering-based and neighbor knowledge schemes [10,23]. So if you choose the forwarding
neighbors, the node selects some of its 1-hop neighbors as rebroadcasting nodes. In the cluster structure, the network is par-
titioned into a group of clusters. Each cluster has one cluster head that controls all other members in the cluster. Gateway
nodes are used for routing between clusters. The rebroadcast is performed by cluster heads and gateways.
Khelil et al. in [24] introduce hyper-gossiping, a novel adaptive broadcast algorithm that combines two strategies. Hyper-
gossiping uses adaptive gossiping to efficiently distribute messages within single network partitions and implements an effi-
cient heuristic to distribute them across partitions. In [25] Tseng et al., also proposed two adaptive heuristic-based schemes,
called adaptive counter-based (ACB) and adaptive location-based (ALB). The authors derived the best appropriate counter
threshold and coverage-threshold for ACB and ALB as a function of the number of neighbors. The authors showed that these
adaptive schemes outperform the non-adaptive schemes and recommend ACB if location information is unavailable and sim-
plicity is required.
3. The proposed scheme
In this section, we present a new counter-based scheme; we will use it to reduce the broadcast storm problem associated
with flooding. The use of this scheme is to enable the mobile nodes to rebroadcast a message if the number of received dupli-
cate packets is less than a threshold by taking in consideration the status of the node density like sparse and dense areas. In
this paper we will begin to use the counter-based scheme to know the amount of delivery.
We begin in the first phase with the new counter-based algorithm using three dynamic thresholds to present a more effi-
cient broadcast solution in sparse and dense networks by initiating the counter cthat will count the number of times which
the node receives the same packet and increment this cfor each same broadcast packet. To determine the three counter
thresholds, you need to calculate through Hello packets and know the node with large number of neighbors (the number
of neighbors is decided to be large or small based on a pre-set threshold value) that will have a high priority to receive
the broadcast first.
Using neighbor’s information which can be calculated through Hello packets, we determine if the node is located within
dense, medium distribution or sparse regions and assigned C
min
,C
mid
, and C
max
thresholds, respectively. Then we need to
calculate the random assessment delay (RAD, which is randomly chosen between 0 and T
max
seconds by dividing random
number between 0 and 1 and Random Factor (RF) such as (RAD = X/RF)).
If the total number of neighbors nis less than the average number of neighbors then the node may exist in sparse area and
it will take the smallest threshold C
max
, but if the number of neighbors nis within the average number of neighbors, then the
node exist in medium area, so it will take the medium threshold C
mid
, or if the number of neighbors nis more than the aver-
age number of neighbors, then the node exist in dense area, so it will take the dense threshold C
min
; otherwise the rebroad-
cast is stopped [13]. We evaluate the performance of our proposed technique using the ns-2 network simulator [26]. Ns-2 is a
well known discrete event simulator used widely in research for both wired and wireless networks. The simulation param-
eters that have been used in our experiments are shown in Table 1. The new algorithm in phase one shows in Fig. 1.
4. Simulation results and analysis
In this section, we present the various simulation experiments that we performed to evaluate the performance of our pro-
posed scheme. The main focus of our simulations is to simulate AODV routing discovery, which is based on using simple
M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563 555

flooding as a route discovery mechanism and simulate the counter-based broadcasting scheme with three counter thresh-
olds and two counter thresholds. In the following sections, we present and analyze the results of the simulation experiments
that aim to compare our new counter-based scheme with two counter-based broadcasting scheme and blind flooding using
different simulation metrics.
4.1. Effects of offered traffic load simulation
Like the previous studies, the offered traffic load simulation is done by changing the number of Constant Bit Rate (CBR)
connections. This CBR connection ensures that all cells in a transmission are maintained from end to end. This service type is
used for voice and video transmission that require little or no cell loss and rigorous timing controls during transmission. The
numbers of CBR connections that are considered in the experiments are 10, 20, 30 and 40 for the number of nodes is 50 and
20, 40, 60 and 80 for the number of nodes is 100. The maximum speed 20 m/s is chosen to study the effects of traffic load in
the network with high speed. When the speed is high the traffic load is concentrated on some nodes so the congestion is
occurred.
The simulation parameters of this experiment are set as follows:
Number of nodes: 50 and 100 nodes.
Maximum speed: 20 m/s.
Packet rate: 4 packets/second.
Number of sources for 50 nodes = 10, 20, 30 and 40 CBR generators.
Number of sources for 100 nodes = 20, 40, 60 and 80 CBR generators.
Node pause time = 0 s.
Protocol receiving ()
1. On hearing a broadcast packet m at node X:
2. Get the Broadcast ID from the message; n1 minimum numbers of neighbor, n2
maximum number of neighbor, all are threshold values;
3. Get degree n of a node X (number of neighbors of node X);
4. If packet m received for the first time then
5. If n < n1 then
6. Node X has a low degree: low threshold value = cmin;
7. Else If n1<= n <= n2 then
8. Node X has a medium degree: medium threshold value = cmid;
9. Else If n> n2 then
10. Node X has a high degree: high threshold value = cmax;
11. End if
12. Timer = 1
13. While (the message not hearing to start the transmission) Do
14. Wait for a random number of slots until the transmission is actually starts.
15. End while
16. Increment the counter-threshold
17. If (counter-threshold< threshold) Go to 13
18. Else exit algorithm
19. End if
Fig. 1. 3C’s threshold of a new counter-based broadcasting scheme.
Table 1
Simulation parameters.
Simulation parameters Parameter value
Simulator used Ns-2 (version 2.29)
Bandwidth 2 Mbps
Packet size 512 bytes
Topology size 500 500 m
2
Number of nodes 25,50,75,100
Mobility model Random waypoint
Simulation time 500 s
MAC protocol IEEE 802.11
Traffic type CBR
Transmission range 100 m
Number of scenario 30 scenarios
556 M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563

4.1.1. Normalized Routing Load (NRL) for 50 nodes
Fig. 2 shows the results of the normalized routing load vs. the network sizes (number of connections) for all the three
schemes. Apparently, this figure shows that increases in connections tend not to lead to noticeable increase in the NRL using
our proposed scheme. When the traffic load increased, there exist many connections between any nodes used to reach to the
destination, so we choose one of these connections. Most of the generated data packets and connections are dropped result-
ing from collisions and contention. Nevertheless, our proposed scheme will decrease the NRL over the traffic load percentage
against other schemes and shows better performance up to 30%. This is because the flooding sends the packets to all nodes
continuously without checking if these nodes receive this packet in previous time; thus this causes a collision and contention
in the network leading to additional load on the network.
4.1.2. Average end-to-end delay for 50 nodes
Fig. 3 represents the delays of all schemes for different traffic loads. The delay is increased as the traffic load grows. The
number of packets transmitted on the network has a considerable impact on delay. When the number of CBR connections
increases the number of collisions, contentions and redundant rebroadcast packets grows. Thus, this leads to more retrans-
missions of packets towards the destination and, hence, resulting in growing delay. Fig. 3 shows that flooding incurs higher
end-to-end delay. This is owing to the higher number of redundant rebroadcasts of RREQ packets with collisions and con-
tention caused by many RREQ packets that fail to reach the destination.
4.1.3. Packet delivery ratio (PDR) for 50 nodes
Fig. 4 represents the PDR for all schemes in this study. This figure shows that our proposed scheme has a higher value of
PDR compared with two counter-based and flooding. Packet delivery ratio increases when increasing the number of connec-
tions for the following reason: the more the network connections, the better and more available shortest paths towards des-
tination. This implies that there are more connections to connect two nodes offering a better transmission in each area.
Hence, there is a greater chance that a broadcast retransmission occurs successfully, resulting in an increased delivery ratio.
1.000
1.300
1.600
1.900
2.200
2.500
2.800
3.100
0 1020304050
Number of Connections
Normalized Routing Load
flooding
2 counters
3 counters
Fig. 2. NRL vs. traffic load of network size = 50 and node speed 20 m/s.
0
0.01
0.02
0.03
0.04
0.05
0.06
01020304050
Number of Connections
End-to-End Delay (sec)
flooding
2 counters
3 counters
Fig. 3. Average delay vs. traffic load of network size = 50 and node speed 20 m/s.
M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563 557

4.1.4. Routing overhead for 50 nodes
Fig. 5 shows that with increasing the traffic load, the total packets sent (either data or control packets) increases and the
number of source nodes that initiate the route discovery operations is also increased, resulting in a larger routing overhead in
all schemes. In addition, as the number of packets being sent increases, the probability that these packets collide becomes
larger, leading to more re-send of these packets, and thus, more route discovery overheads. For both low number and high
number of connections, our proposed scheme outperforms the other schemes.
These figures illustrate that our scheme achieves better performance in terms of normalized routing load, average end-to-
end delay, packet delivery ratio and routing overhead and survives under heavy load. Table 2 presents the percentages of
improvements that our scheme achieved over other schemes at 10, 20, 30 and 40 CBR generators of 50 nodes.
4.1.5. Normalized Routing Load (NRL) for 100 nodes
Fig. 6 plots the results of the normalized routing load vs. the traffic load for all three schemes. It can be noticed from the
figure that the NRL grows when the traffic load increases. However, our proposed scheme exhibits the lowest traffic load
percentage against other schemes. This is because the flooding sends the packets to all nodes continuously without checking
92
93
94
95
96
97
98
0 1020304050
Number of Connections
Pd Fraction (%)
flooding
2 counters
3 counters
Fig. 4. Packet delivery ratio vs. traffic load of network size = 50 and node speed 20 m/s.
50200
50700
51200
51700
52200
52700
0 1020304050
Number of Connections
Routing Overhead (RREQ packets)
flooding
2 counters
3 counters
Fig. 5. Routing overhead vs. traffic load of network size = 50 and node speed 20 m/s.
Table 2
Improvement percentages of our scheme over two counter-based and blind flooding at traffic load simulation of 50 nodes.
Metrics/number of connections 10 connections 20 connections 30 connections 40 connections
Flooding (%) 2C’s (%) Flooding (%) 2C’s (%) Flooding (%) 2C’s (%) Flooding (%) 2C’s (%)
Normalized routing load 27 11 30 16 23 12 25 14
Average delay 23 10 20 7 20 11 18 9
Packet delivery ratio 5 2 4.5 1 4 1.5 2.6 1
Routing overhead 3 2 4 2 3 2 5 2
558 M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563

if these nodes receive the packets previously; thus leading to significant collisions, contentions in the network and additional
load.
4.1.6. Average end-to-end delay for 100 nodes
Fig. 7 illustrates the delays of all schemes for different traffic loads of network size 100. The delay grows proportional to
the volume of traffic load. This figure shows that flooding has higher end-to-end delay value as a result of the higher number
of redundant rebroadcasts of RREQ packets with collisions and contentions in which many RREQ packets fail to reach the
destination, resulting in increasing delay.
4.1.7. Packet delivery ratio (PDR) for 100 nodes
Fig. 8 represents the PDR for the all schemes. This figure shows that our proposed scheme has a higher value of PDR com-
pared with both, two counter-based and flooding. Packet delivery ratio grows when increasing the number of connections.
This implies that there are more connections to connect two nodes and facilitate the transmission in each area. Hence, there
is a greater chance that a broadcast retransmission occurs successfully, resulting in an increased delivery ratio.
4.1.8. Routing overhead for 100 nodes
Fig. 9 shows that with increasing the traffic load, the total packets sent (either data or control packets) increase, resulting
in a larger routing overhead in all schemes. In addition, as the number of packets being sent (i.e. the traffic load) increases,
the probability that these packets collide becomes larger, leading to more re-send cases for these packets, and thus, more
route discovery overheads incurred. As shown in Fig. 9, for low number of connections of 10 connections, our proposed
scheme outperforms the other schemes.
These figures illustrate that our scheme achieves better performance in terms of normalized routing load, average end-to-
end delay, packet delivery ratio and routing overhead and survives under heavy load traffic. Table 3 presents the percentages
of improvements that our scheme achieved over its counterparts at 20, 40, 60 and 80 CBR generators of 100 nodes.
25
28
31
34
37
40
020406080100
Number of Connections
Normalized Routing Load
flooding
2 counters
3 counters
Fig. 6. NRL vs. traffic load of network size = 100 and node speed 20 m/s.
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
020406080100
Number of Connections
End-to-End Delay (sec)
flooding
2 counters
3 counters
Fig. 7. Average delay vs. traffic load of network size = 100 and node speed 20 m/s.
M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563 559

4.2. Effects of node mobility simulation
The purpose of the simulations presented in this subsection is to study the effects of different speeds of nodes, within a
very large network density, on the performance of the three examined protocols. In all the experimental scenarios, the max-
imum node speed is varied from 1 m/s up to 20 m/s, i.e., the experiments study the effect of very low node speed (min
speed = 1 m/s) and very high speed (max speed = 20 m/s). By assuming that each node in the network has a transmission
range of 250 m, we consider the network that has 50 nodes each of which has 250 m transmission range. The results of Figs.
10–13 illustrate the NRL, average end-to-end delay, packet delivery ratio, and routing overhead, respectively. The simulation
parameters of this experiment are set as follows:
Number of nodes: 50 nodes.
Maximum speed: 1, 5, 10, and 20 m/s.
97.200
97.300
97.400
97.500
97.600
97.700
97.800
0 20406080100
Number of Connections
PD fraction (%)
flooding
2 counters
3 counters
Fig. 8. Packet delivery ratio vs. traffic load of network size = 100 and node speed 20 m/s.
50900
51000
51100
51200
51300
51400
51500
0 20406080100
Number of Connections
Routing Overhead (RREQ packets)
flooding
2 counters
3 counters
Fig. 9. Routing overhead vs. traffic load of network size = 100 and node speed 20 m/s.
Table 3
Four improvement percentages of our scheme over two counter-based and blind flooding at traffic load simulation of 80 nodes.
Metrics/number of connections 20 connections 40 connections 60 connections 80 connections
Flooding (%) 2C’s (%) Flooding (%) 2C’s (%) Flooding (%) 2C’s (%) Flooding (%) 2C’s (%)
Normalized routing load 10 8.5 9 8.5 9 8.3 10.5 8.3
Average delay 16 10 13 8 13 8 20 11
Packet delivery ratio 1 0.8 1 0.9 1 0.4 1 0.4
Routing overhead 2 2 3 2 3 2 2 2
560 M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563

Packet rate: 4 packets/s.
Number of sources = 40 CBR generators.
Node pause time = 0 s.
4.2.1. Normalized Routing Load (NRL) for 100 nodes
Fig. 10 shows the normalized routing load achieved by the three schemes with different speeds. This figure shows that the
NRL increases proportional to the node mobility. This is due to the fact that when the node mobility grows, the probability of
24
25
26
27
28
29
30
0 5 10 15 20 25
Node Speed (m/sec)
Normalized Routing Load
flooding
2 counters
3 counters
Fig. 10. Normalized routing load vs. node speed of network size = 50.
0.00
0.02
0.04
0.06
0.08
0.10
0 5 10 15 20 25
Node Speed (m/sec)
End-to-End Delay (sec)
flooding
2 counters
3 counters
Fig. 11. End-to-end delay vs. node speed of network size = 50.
98.600
98.620
98.640
98.660
98.680
98.700
0 5 10 15 20 25
Node Speed (m/sec)
Pd Fraction (%)
flooding
2 counters
3 counters
Fig. 12. Packet delivery ratio vs. node speed of network size = 50.
M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563 561

network topology changes also increases and the nodes can be quickly connected to transmit the data packets where large
numbers of RREQ packets are generated to transmit packets to the destination. In our proposed scheme, NRL becomes evi-
dently smaller by 11%. This is due to the fact that our scheme uses the counter thresholds to determine the sparse, medium
and dense area so as to reduce the number of RREQ generated packets that need to be delivered to the destination.
4.2.2. Average end-to-end delay
Fig. 11 displays the average end-to-end delay for the three schemes when the number of nodes is 50. The faster the node
speed is, the higher end-to-end delay is incurred. The results show that for the average value of node speeds, our scheme are
40% better than both two counter-based and blind flooding. When the speed is low, the new scheme outperforms other
schemes by 20%. At high speed values, our scheme outperforms its counterparts by 40%.
4.2.3. Packet delivery ratio (PDR)
Fig. 12 depicts the delivery ratio for the three examined schemes at different speeds when network density is set at 50
nodes. The figure shows that our proposed scheme outperforms other schemes at all speed values. As can be noticed, the
delivery ratio achieved by the three protocols decreases as nodes speed increases. This behavior is logical since the faster
the nodes are, the less stable links can be offered, resulting in lower delivery ratio.
4.2.4. Routing overhead
Fig. 13 shows that for all different speeds and with more mobile nodes in the network, the routing overhead becomes
evidently larger. The overhead encountered by three counter-based, two counter-based and blind flooding increases by
30%, 19%, and 10%, respectively. This figure shows that with the increasing maximum speed of nodes, the total packets sent
(either data or control packets) increases, resulting in a larger routing overhead. In addition, as the number of packets being
sent increases, the probability that these packets will collide becomes larger. As a consequence, this leads to more re-send
packets, and thus, more route discovery overheads as depicted in Fig. 13.Table 4 presents the percentages of improvements
that our scheme achieved.
5. Conclusion
The new proposed counter-based broadcasting scheme is a reliable scheme that avoids the ‘‘brute force” attitude of
simple flooding which causes very high overhead, especially at large dense networks. The main goal of our scheme is to re-
duce the overhead and delay resulting from flooding and maximize the packet delivery ratio and the NRL. By using this new
0
2000
4000
6000
8000
10000
12000
14000
16000
0 5 10 15 20 25
Node Speed (m/sec)
Routing Overhead (RREQ packets)
flooding
2 counters
3 counters
Fig. 13. Routing overhead vs. node speed of network size = 50.
Table 4
Improvement percentages of our scheme over two counter-based and blind flooding at node speed simulation.
Metrics/node speed 1 m/s 5 m/s 10 m/s 20 m/s
Flooding (%) 2C’s (%) Flooding (%) 2C’s (%) Flooding (%) 2C’s (%) Flooding (%) 2C’s (%)
Normalized routing load 11 8 11 8 11 8 11 8
Average delay 17 13.8 20.3 14.3 23 17 40 19
Packet delivery ratio 1 0.5 1 0.9 1 0.4 1 0.4
Routing overhead 7 5 10 4 11 3 19 8
562 M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563

mechanism, our scheme introduces a very low overhead compared to existing schemes used in this study. The simulation
experiments show that our proposed scheme significantly outperforms the two counter-based and flooding schemes in
terms of reducing overhead by 28% for low speed of nodes (1 m/s), and by 58% for high speed of nodes (20 m/s). In addition,
a new counter-based scheme substantially outperforms others in terms of reducing average end-to-end delay by 48% for low
speed and by 56% for high speeds. Pertaining to the packet delivery ratio, the experiments show that our scheme outper-
forms others by 4%. Finally, our scheme achieves substantial improvement of the NRL by 23%. As a future work, we plan
to implement our scheme in next generation ad hoc and vehicular networks.
References
[1] C.S.R. Murthy, B.S. Manoj, Ad Hoc Wireless Networks: Architectures and Protocols (2004) 2–5.
[2] M. Aminu, M. Ould-Khaoua, L.M. Mackenzie, J. Abdulai, Performance evaluation of an efficient counter-based scheme for mobile ad hoc networks based
on realistic mobility model, Proceedings of IEEE International Conference on Signal Processing and Communications (2009) 1–4.
[3] Y.-C. Tseng, S.-Y. Ni, E.-Y. Shih, Adaptive approaches to relieving broadcast storms in a wireless multihop mobile ad hoc network, in: Proceedings of the
21st International Conference on Distributed Computing Systems, 2001, pp. 481–488.
[4] M. Bani-Yassein, M. Ould-Khaoua, L.M. Mackenzie, S. Papanastasiou, Performance analysis of adjusted probabilistic broadcasting in mobile ad hoc
networks, in: Proceedings of the International Journal of Wireless Information Networks, 2006, pp. 1–14.
[5] S.-Y. Ni, Y.-C. Tseng, Y.-S. Chen, J.-P. Sheu, The broadcast storm problem in a mobile ad hoc network, in: Proceedings of ACM/IEEE Mobicom ’99, 1999,
pp. 151–162.
[6] C. Perkins, E.M. Royer, Ad-hoc on-demand distance vector routing, Proceedings of 2nd IEEE Workshop on Mobile Computing Systems and Applications
(1999) 90–100.
[7] V.D. Park, S. Corson, Temporally-ordered routing algorithm (TORA) version 1: functional specification, corson-draft-ietf-manet.tora-spec-00.txt, IETF,
Internet draft, 1997. Available from: <http://tools.ietf.org/html/draft-ietf-manet-tora-spec-00>.
[8] C. Ho, K. Obraczka, G. Tsudik, K. Viswanath, Flooding for reliable multicast in multihop ad hoc networks, Proceedings of ACM (1999) 64–71.
[9] R. Khurram, A survey of mobile ad hoc networks, in: Proceedings of the 2nd IEEE Workshop on Mobile Computing Systems and Applications, May 2002.
[10] S.-Y. Ni, Y.-C. Tseng, Y.-S. Chen, J.-P. Sheu, The broadcast storm problem in a mobile ad hoc network, Wireless Networks (2002) 153–167.
[11] J. Cartigny, D. Simplot, Border node retransmission based probabilistic broadcast protocols in ad-hoc networks, in: Proceedings of HICSS-36, 2003.
[12] A. Mohammed, M. Ould-Khaoua, L. Mackenzie, J. Abdulai, Improving the performance of counter-based broadcast scheme for mobile ad hoc networks,
in: Proceedings of 2007 IEEE International Conference on Signal Processing and Communications, 2007, pp. 1403–1406.
[13] A. Mohammed, M. Ould-Khaoua, L. Mackenzie, An efficient counter-based broadcast scheme for mobile ad hoc networks, in: Proceedings of the Fourth
European Performance Engineering Workshop Lecture Notes in Computing Science, 2007, pp. 275–283.
[14] A. Qayyum, L. Viennot, A. Laouiti, Multipoint relaying for flooding broadcast messages in mobile wireless networks, in: Proceedings of the 35th Annual
Hawaii International Conference on System Sciences, 2002, pp. 60–65.
[15] G. Calinescu, I. Mandoiu, P.-J. Wan, A.Z. Zelikovsky, Selecting forwarding neighbors in wireless ad hoc networks, Mobile Networks and Applications 9
(2004) 101–111.
[16] I. Stojmenovic, M. Seddigh, J. Zunic, Dominating sets and neighbor elimination-based broadcasting algorithms in wireless networks, Proceedings of
IEEE (2002) 14–25.
[17] W. Lou, J. Wu, A cluster-based backbone infrastructure for broadcasting in MANETs, in: Proceedings of IEEE Computer Society, 2003.
[18] Y.-C. Tseng, S.-Y. Ni, En-Y.U. Shih, Adaptive approaches to relieving broadcast storm in a wireless multihop mobile ad hoc network, IEEE Transactions
on Computers (2003) 545–557.
[19] B. Williams, T. Camp, Comparison of broadcasting techniques for mobile ad hoc networks, in: Proceeding of MOBIHOC, ACM, 2002. pp. 194–205.
[20] W. Peng, X.C. Lu, The reduction of broadcast redundancy in mobile ad hoc networks, Proceedings of the First Annual Workshop on Mobile Ad Hoc
Networking & Computing (2000) 129–130.
[21] Q. Zhang, D.P. Agrawal, Dynamic probabilistic broadcasting in MANETs, Proceedings of Journal of Parallel and Distributed Computing (2005) 220–233.
[22] J. Wu, W. Lou, Forward-node-set-based broadcast in clustered mobile ad hoc networks, in: Proceedings of the Wireless Networks and Mobile
Computing, 2003, pp. 155–173 (special issue on Algorithmic, Geometric, Graph, Combinatorial, and Vectors).
[23] M. Bani-Yassein, M. Ould-Khaoua, L.M. Mackenzie, S. Papanastasiou, A. Jamal, Improving route discovery in on demand routing protocols using local
topology information in MANETs, in: Proceedings of the Ninth ACM/IEEE International Symposium on Modeling, Analysis and Simulation of Wireless
and Mobile Systems, 2006, pp. 95–99.
[24] A. Khelil, P. Jose Marron, Ch. Becker, K. Rothermel, Hypergossiping: a generalized broadcast strategy for mobile ad hoc networks, Ad Hoc Networks5
(2007) 531–546.
[25] Y.C. Tseng, S.Y. Ni, E.Y. Shih, Adaptive approaches to relieving broadcast storms in a wireless multihop mobile ad hoc networks, Proceedings of the IEEE
Transactions on Computers (2003).
[26] The Network Simulator ns-2. <www.isi.edu/nsnam/ns>.
M.B. Yassein et al. / Simulation Modelling Practice and Theory 19 (2011) 553–563 563