INTERNATIONAL JOURNAL OF COMMUNICATIONS

Issue 1, Volume 2, 2008

Comparative Performance Study of ADMR and

ODMRP in the context of Mobile Ad Hoc
Networks and Wireless Sensor Networks
Radosveta Sokullu , Ozlem Karaca

sector where vital patient information is collected by wireless

Abstract— Mobile Ad Hoc networks (MANET) and Wireless sensors and transmitted to only interested or responsible
Sensor Networks (WSNs) are two large groups of wireless networks personnel (doctors, nurses involved with a certain patient [1]),
that have well established application ranges. Despite the fact that tracking of fire-fighters in burning buildings, data collection
they address very distinctive groups of devices and have clearly
with mobile sensors, disaster rescue etc. These scenarios
differentiated wireless interfaces, there are certain similarities which
push scientists to look for adopting solutions already designed for require more general topologies than the event-to-sink model
existing wireless networks to WSNs. An example of this is the case usually accepted for WSNs. When comparing the two
with routing layer protocols. AODV, a unicast routing protocol, protocols the underlying media access mechanism has been
developed for MANETs, has proved to be applicable and was taken into consideration and IEEE 802.11 and IEEE 802.15.4
accepted by IEEE as the standard for the routing layer in Low Rate – have been covered.
Wireless Personal Area Networks (LR-WPAN). MANET-originated
Performance comparison research has been done before for
solutions, like multicast protocols, have also been initially designed in
the context of IEEE 802.11 MAC layer protocol but have their multicast protocols based on IEEE 802.11 MAC layer [13]. In
applicability for WSNs not been studied so far. This paper [14] a comparison is presented for IEEE 802.11 and 802.15.4
investigates the feasibility of two popular MANET multicast using AODV at the routing layer. The effect of using an
protocols, ADMR and ODMRP over the IEEE 802.15.4 standard and RTC/CTS mechanism on the packet delivery ratio is
provides a comprehensive study of the performance of these two investigated and it is proved that even in collision free
protocols with different underlying physical and media access environments the ratio of RTS/CTS packets to the data packets
protocols. The protocols have been analyzed with ns-2 network
simulator. It appears that even though both protocols are applicable in is quite high because they are also used for transmissions of
the selected scenarios, there are specifics in their performance in the control packets of the network layer. In this work ADMR and
context of WSNs which should not be neglected. ODMRP were selected representing two different groups of
multicasting, with two different underlying MAC layer
Keywords— Wireless LAN, Mobile Ad Hoc Networks, wireless protocols, respectively IEEE 802.11 and 802.15.4. The
sensor networks, medium access control mechanisms, multicast relationship between network protocols and MAC layer
routing protocols, performance evaluation . protocols is investigated in diverse scenarios based on the
following parameters: packet delivery ratio (PDR), protocol
I. INTRODUCTION overhead and effects of mobility.
Wireless sensor networks (WSNs) enhanced with actuator The paper is structured as follows: the next two sections
capabilities materialize the interface between people and the provide a brief background on the specifics of the protocols
environment and establish a context for assisted living and that are investigated, first for the medium access control and
emergency measures, intelligent production and transport, and then for the routing layer. In Section IV the simulation model
environmental monitoring. Existing solutions in different OSI and the methodology use is discussed. In Section V the
layers, designed initially for MANETS, are tested for their simulation results are presented followed by conclusions in
applicability in WSNs. An example is the adoption of AODV Section VI, which summarize the most important contributions
as a routing protocol for LR-WPAN. The focus of this paper is of the work.
further investigating such solutions, like ADMR and ODMRP,
which are multicast protocols originally designed for II. SPECIFICS OF THE IEEE 802.11 AND IEEE 802.15.4 MAC
MANETs, in the context of WSN application scenarios and LAYER PROTOCOLS
performance requirements. An open question is whether the Both protocols have been standardized by IEEE for the
multicast supporting functions of routing protocols developed physical (PHY) and media access control (MAC) layer of
for MANETS like ADMR and ODMRP can be used for wireless networks but aiming at different types of wireless
WSNs. Need for such functions has been seen in many WSN devices and network configurations.
based application scenarios like in the health sector where vital The 802.11 addresses wireless networks consisting of
laptops or similar class of devices, in either infrastructure or

45 Manuscript Received January 17, 2008; Revised Received March 24, 2008
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

infrastructure-less (Ad Hoc) mode. IEEE 802.11 series When a transmitter is ready for transmission of a frame, it
standards have been widely used as the MAC layer protocol in checks the status of the channel. If the channel is busy, it waits
wireless networks, which specify the arbitration of channel until the end of the transmission progress. This part makes the
access under contentions among multiple wireless transmission DCF a CSMA protocol. When the channel becomes idle,
devices. In particular, the IEEE 802.11a/b/g standards are used rather than transmitting immediately, the transmitter selects a
to specify the MAC mechanism in wireless local area networks random backoff interval to reduce the collision probability.
(WLANs). This part makes the DCF a collision avoidance protocol. If a
Whereas IEEE 802.11 are interested in features such as data frame is successfully received, the receiver replies with an
Ethernet matching speed, long range (100m), complexity to ACK (Acknowledgement) frame, after a SIFS (Short Inter-
handle seamless roaming, message forwarding, and data Frame Space) time interval. It is only after receiving an ACK
throughput of 2-11Mbps, WPANs are focused on a space frame correctly that the transmitter assumes a successful
around a person or object that typically extends up to 10m in delivery of the corresponding data frame. On the other hand, if
all directions [4]. The focus of WPANs is low-cost, low an ACK frame is received in error or no ACK frame is
power, short range and very small size devices. The IEEE received, this is meaning that a failure of the corresponding
802.15 WG has currently defined three classes of WPANs – data frame transmission occurs. Fig. 1 illustrates frame
802.15.1 (Bluetooth), 802.15.4 (ZigBee), 802.15.3 (UWB) - exchange sequences of the DCF [3].
which are distinguished by data rate, battery drain and quality In multi-hop environments, a frame exchange sequence with
of service (QoS). The low rate WPANs (IEEE 802.15.4/LR- an RTS/CTS (Request-to-Send/Clear-to-Send) exchange is
WPAN) are intended to serve a set of industrial, residential used to mitigate the hidden/exposed node problems. Nodes
and medical applications with very low power consumption, that overhear the duration field of RTS/CTS set their NAV
cost requirement and with relaxed needs for data rate and QoS (Network Allocation Vector) in order not to interrupt a data
which include wireless sensor nodes as well. The low data rate transmission following the RTS/CTS exchange. The SIFS,
enables the LR-WPAN to consume very little power. which is the smallest time interval used between two
The paper concentrates on the performance comparison of consecutive frame transmissions, is used within this four-way
two different multicast routing protocols, originally suggested – RTS-CTS-data-ACK – handshake. Other nodes must wait
for Ad Hoc networks, ADMR and ODMRP, using different for an idle medium for at least DIFS (DCF Inter-Frame Space)
underlying MAC layer protocols, specifically IEEE 802.11 and time interval, and hence are prevented from attempting to use
IEEE 802.15.4. It is accepted that the protocols designed in the medium.
accordance with the OSI network model should be
independent from the underlying layer. Even though this
statement is true in general, it is interesting to investigate if
there are any specifics in the performance related to the
different mechanisms of accessing the media and formulate the
conditions for the applicability of MANET-originated
solutions to WSN. Fig. 1 802.11 DCF.
802.11 WLAN technology specifies both the Medium
Access Control (MAC) and the Physical Layers (PHY). The The IEEE 802.15.4, covering the PHY and MAC layer, is
802.11 WLAN PHY layer is responsible for the selection of developed for LR-WPANs, providing ad hoc self-organizing
the correct modulation scheme given the channel conditions functionality among inexpensive fixed, portable and moving
and provides the necessary bandwidth. This standard allows devices for applications with relaxed throughput requirements
the same MAC layer to operate on top of one of several PHY [5].
layers. The difference among 802.11/a/b/g/n WLAN standards The PHY layer of the IEEE 802.15.4 provides two services:
is mainly on carrier frequencies and on transmission speed. the PHY data service and PHY management service
The MAC layer of IEEE 802.11 decides in a distributed interfacing to the physical layer management entity (PLME).
manner on how the offered bandwidth is shared among all The PHY data service makes possible the transmission and
stations to provide wireless connectivity. Fairness and reception of PHY protocol data units (PPDU) across the
maximum bandwidth utilization are a major design goal. Two physical radio channel. The features of the PHY are activation
forms of MAC layer have been defined in IEEE 802.11 and deactivation of the radio transceiver, energy detection
specification named, Distributed Coordination Function (DCF) (ED), link quality indication (LQI), channel selection, clear
and Point Coordination Function (PCF). The DCF protocol channel assessment (CCA) and transmitting in addition to
uses Carrier Sense Multiple Access with Collision Avoidance receiving packets across the physical medium. The PHY
(CSMA/CA) and is mandatory, while PCF is defined as an among other functions specifies the receiver sensitivities as -
option to support time-bounded delivery of data frames [2]. 85dBm for 2.4GHz and -92dBm for 868/915MHz. The
The DCF protocol combines the carrier sensing with RTS/CTS achievable range is a function of the receiver sensitivity and
handshake to reduce interference and cope with the hidden the transmit power.
terminal problem. The MAC sublayer provides two services: the MAC data

46
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

service and the MAC management service interfacing to the In a non beacon-enabled network
MAC sublayer management entity (MLME) service access
point (SAP). The MAC data service enables the transmission III. AD HOC MULTICAST ROUTING PROTOCOLS FUNCTIONAL
and reception of MAC protocol data units (MPDU) across the OVERVIEW ATH
PHY data service. Besides the non-beacon-enabled mode, the The current Internet routing system relies primarily on two
MAC sublayer is capable of supporting beacon management in basic algorithms and their variations. Link-state routing uses
the beacon-enabled mode, as well as channel access, GTS the Dijkstra algorithm. Distance-Vector routing (e.g., RIP) and
management, frame validation, acknowledged frame delivery, Path-Vector routing (e.g., BGP) use the Bellman-Ford
association and disassociation. There are three types of data algorithm. The multicast routing was developed to enable one-
transfer specified: from a device to a coordinator, from a to-many data delivery in different networks [6]. Multicast is
coordinator to a device and between two peer devices. They the function of transmitting information to a group of nodes
differ depending on the use of beacons. Details on the identified by a single destination address. It has been
communication with the coordinator in beacon-enabled and extensively covered for MANETs. Multicast in WSN has come
non-beacon enabled mode are given in Fig. 2 and Fig.3. up very recently with the emerging of new application
The main functions of the IEEE 802.15.4 MAC layer are scenarios. Providing multicast can greatly reduce the number
grouped based on the use or not of beacons. For the non- of transmitted packets and reduce sensor nodes’ energy
beacon mode, which is investigated in this work, they include consumption because radio transmission is the most power-
channel access (CA), frame validation and acknowledged consuming operation.
frame delivery. The medium access method used is unslotted Unlike the unicast routing protocols, multicast routing
CSMA-CA. A device maintains two variables for each protocols set up routing trees, with their leaves being the end
transmission attempt: NB and BE. NB, is the number of times users in a specific multicast group. This routing tree can be
the CSMA-CA algorithm was required to backoff while initiated either by a source or by receiver. For the source
attempting the current transmission. BE shows how many initiated routing protocols, flooding is used to find the
backoff periods a device must wait before attempting to assess interested users and routers along the path. Based on that
the channel. Although the receiver of the device is enabled multicast routing tables are established by the following
during CA, during that time frame all frames are discarded. multicast packets. Routers without attached interested users
The MAC layer creates delay for a random number of will prune the flooded packets to prevent forwarding of
complete backoff periods in the range 0 to 2BE−1 and then multicast packets. On the other hand, a receiver can also
requests PHY to perform a CCA. initiate a multicast routing tree by sending a join message to
the source and the response from the source will enable the
routers on path to establish the multicast routing tables.
Multicast routing protocols can be divided into two categories:
tree-based and mesh-based according to how packets are
routed through the network [7].
Like in “wired" multicast routing, tree-based protocols build
a tree over which multicast data is forwarded. Since some of
MANETs' key features, like fast deployment, make them
attractive for deployment in critical environments, such as
military or civilian emergency operations, robustness and
reliability are essential. Thus, one of the main challenges faced
by multicast routing in MANETs is the need to achieve
robustness in the presence of universal mobility and frequent
Fig. 2 Communication to a coordinator
In a beacon-enabled network node outages and failures. In a tree-based paradigm data is
propagated over a spanning tree connecting all multicast group
members while mesh-based ones forward data to all group
members over a subset of the nodes. Mesh-based routing
builds a mesh over which multicast data is forwarded and thus
addresses MANET's robustness requirements through data
path redundancy [8].
Our study compares the performance of the On-Demand
Multicast Routing Protocol (ODMRP) [9] as the representative
of mesh-based protocols against Adaptive Demand Driven
Multicast Routing Protocol (ADMR) [10] representing tree-
based schemes.
Fig. 3 Communication to a coordinator

47
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

A. On Demand Multicast Routing Protocol packets are still present.

In ODMRP [9], group membership and multicast routes are A node which receives a multicast packet forwards it only if
established and updated by the source on demand. A request this is a non-duplicate packet. It also sets a flag to show that
phase and a reply phase comprise the protocol like in on- this multicast group is already in use and not expired. Besides
demand unicast routing protocols. A source node that has minimizing the traffic overhead, this procedure prohibits
packets to send broadcasts an advertising packet, JOIN sending packets through stale routes.
QUERY to the whole network. This periodic transmission In ODMRP, no explicit control packets need to be sent to
refreshes the membership information and updates the route. join or leave the group. If a multicast source wants to leave the
ODMRP does not maintain route information permanently. It group, it simply stops sending JOIN REQUEST packets since
uses a soft state approach in group maintenance. When an it does not have any multicast data to send to the group.
intermediate node receives a non-duplicate JOIN QUERY, it There are three types of tables in ODMRP architecture:
stores the upstream node’s ID in its "Message Cache" in Member Table, Routing Table and Forwarding Group Table.
order to use this information later for transmission in backward The Member Table is used for storing the source information.
direction and rebroadcasts the packet. The upstream node Each entry is designated by “source ID” and “time of last JOIN
address is inserted or updated as the next node for the source REQUEST received” pair. If no JOIN REQUEST is received
node in its "Routing Table." If the JOIN QUERY packet is within a refresh period, that entry is removed. The Routing
not a duplicate and the Time-To-Live value is greater than Table is created on demand and is maintained by each node. It
zero, appropriate fields are updated and it is rebroadcast. is updated when a non-duplicate JOIN REQUEST is received.
A cooperating node which performs forwarding, maintains the
When a node receives a JOIN REPLY packet, it checks if group information in the Forwarding Group Table.
the next node ID of one of the entries matches its own ID. If it
matches, the node realizes that it is a part of the forwarding B. Adaptive Demand Driven Multicasting
group of nodes. The nodes which are part of the forwarding
ADMR [11] is an on-demand protocol, thus it does not
group broadcast their own JOIN REPLY packets built upon
maintain route information regularly. Member nodes that
matched entries. Thus the JOIN REPLY is propagated from
constitute the tree are refreshed as needed and do not send
the receiver to the source along the shortest path. This process
explicit leave messages.
forms a mesh of nodes that constitutes the routes between
In ADMR, group membership and multicast routes are
sources and receivers. Multicast senders refresh the
established and updated by the source on demand. Multicast
membership information and update the routes by sending
senders and receivers using ADMR cooperate to establish and
JOIN REPLY packets periodically.
maintain forwarding state in the network to allow multicast
communication.
The multicast forwarding state for a given multicast group G
and sender S in ADMR is conceptually represented as a
loosely-structured multicast forwarding tree rooted at S. Each
multicast packet is dynamically forwarded from S along the
shortest-delay path through the tree to the receiver members of
the multicast group.
The forwarding tree consist of receivers, sources and
forwarding nodes that they are not receivers for this group but
only their duties are packet forwarding along the path. All
mentioned nodes constituting the tree are member nodes. Only
members of the multicast forwarding tree forward multicast
packets, and each node forwards each packet at most once.
Duplicate packet suppression is supplied by the protocol by
means of created routing tables.

Fig. 4 Mesh formation in ODMRP

After the establishment of the groups and the route

construction process, a multicast source can transmit packets to
receivers via selected routes and forwarding groups. Periodic
control packets are sent only when outgoing data packets are

48
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

Fig. 5 Network and Tree flood in ADMR

When multicast sources have packets to send, but do not Fig. 6 Multicast State Setup
have routing or membership information, they flood an
advertisement packet to all nodes in the network, which is
known as a network flood. The flood of a packet constrained
to the nodes in the multicast forwarding tree is known as a tree
flood, and to the more general type of flood of a packet
through all nodes as a network flood. (Fig. 5).
This use of flooding within the multicast forwarding tree is
similar to the “forwarding group” concept introduced in the
ODMRP, except that the forwarding state of ADMR is specific
to each sender rather than being shared for the entire group.
When a sender using ADMR sends a multicast packet, it
floods it within the multicast distribution tree only towards the
group’s receivers, whereas with ODMRP, the packet also
floods back towards any other senders that are not receivers Fig. 7 Concluding tree formation
for this group.
The formation of a forwarding tree starts with the flooding Each forwarder or receiver for some multicast group detects
of an advertisement packet called SOURCE INFORMATION that it has become disconnected from the multicast forwarding
packet. A source node sends this packet to produce a Sender tree when it fails to receive a number of successive expected
Table. A Sender Table consists of a “Source ID – S” and multicast data (or keepalive) packets (e.g., 3) from S for G. If
“Group ID – G” touple. Each entry is designated by this this situation occurs REPAIR NOTIFICATION,
touple. When a node receives a SOURCE INFORMATION RECONNECT, RECONNECT REPLY packets are used for
packet, stores the information in its Node Table and the packet finding a new route in the network.
ID to differentiate between trees, and rebroadcasts the packet
to its neighbors. They in turn forward the packet to their IV. SIMULATION MODEL AND METHODOLOGY
neighbors, thus packet floods the whole network. When this Simulation is carried out using ns-2.30 [12]. The simulation
packet reaches a multicast receiver, it is used in the creation of model is based on many-to-many communication model. The
a Membership Table and the broadcasted to the neighbors. At PDR is studied as a function of the node density, the node
the same time the receiver node creates a reply packet called mobility and the varying number of senders and receivers in
MULTICAST SOLICITATION. This packet is also sent in a the network. Overhead is evaluated in respect to the network
broadcast manner. The source node propagates a new type of size.
packet called UNICAST KEEPALIVE packet. It is sent in a
unicast manner from the source to the receiver to reinforce the A. Channel and Radio Model
shortest path. The receiver node upon receiving this UNICAST To accurately model the attenuation of radio waves between
KEEPALIVE packet sends a RECEIVER JOIN packet. As the antennas close to the ground, a model is typically used that
UNICAST KEEPALIVE the RECIEVER JOIN packet is sent attenuates the power of a signal as 1=r2 at short distances (r is
unicast. The tree formation is illustrated in Fig 6. and Fig. 7. the distance between the antennas), and as 1=r4 at longer
distances (Fig.8a,b). The crossover point is called the reference
distance, and is typically around 100 meters for outdoor low-
gain antennas 1.5m above the ground plane operating in the 1–
2GHz band [15]. A two ray ground propagation model is used
in the experiments. In this model, the shadowing fading factor
is not considered. Therefore, for a certain distance, the Pr

49
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

certain distance, the Pr (power at the receiver side) is a the effect of node speed on the packet delivery ratio and on the
deterministic value: incurred overhead in scenarios with different number of sender
S and receiver R nodes.
Pr = Pt Gt Gr ht2 hr2 / L d4 C. Traffic Pattern
A traffic generator was developed to simulate constant bit
In the simulations the Pt_ and the thresholds were adjusted
rate sources. The packet rate is 1 packet per second in all
to set the transmit range to 25 meters for the IEEE 802.15.4
simulations and the size of data payload is 512 bits. The
and 250m for the IEEE 802.11. The CSThresh is set to 1.559e-
senders are chosen randomly among nodes in the network.
11W, RXThresh 3.652e-10 for 802.11 and both to 3.07645e-
Nodes join the multicast session at the time defined by
07W for 802.15.4.
randomly generated traffic scenario and remain so throughout
the simulation.
D. Considered Metrics
The metrics used for the comparison are described in detail
below. Some of them were suggested by the IETF MANET
WG for routing protocol evaluation.
Packet Delivery Ratio (PDR): Determined as the ratio of
the number of data packets actually delivered to the
destinations to the number of data packets supposed to be
received.
Fig. 8a Propagation models- Free Space Model Overhead ratio (OR): Shows the efficiency in terms of
channel utilization and is very important especially in sensor
networks. It is calculated as:
OR = 1 – (Pdata packets sent / Ptotal packets sent) where P
is the number of each type of packets sent by the source node.

V. SIMULATION RESULTS

A. Node density
In this experiment the effect of node density on the PDR is
studied. The number of static nodes varies from 10 to 50 with
1S and 1 or 3R. The results for the different routing protocols
Fig. 8b Propagation models- Two-Ray Model
with IEEE 802.11 and 802.15.4 are given in Fig.9 and Fig.10
respectively.
B. Mobility and Random Way-point Model It is immediately evident that while the PDR is quite stable
The mobility model determines how nodes choose for IEEE 802.11 for the whole range of node densities it is not
destinations for their movement, the speed at which they move, so for the case of IEEE 802.15.4. For densities below 0.005
and the physical paths they take. In the Random Waypoint nodes/m2 the PDR for ODMRP is unacceptably low. ADMR
Mobility (RWP) model, each mobile node begins at a random performs much better. This observation comes to support the
location and moves independently during the simulation. Each thesis made in [9] that a large proportion of control packets
node remains stationary for a specified period that called the required by the network layer protocol even when no
pause time and then moves in a straight line to some new RTS/CTS packets are used, greatly reduces the throughput. For
randomly chosen location at a randomly chosen speed up to densities above 0.005 the performance is quite stable and
some maximum speed. Once reaching that new location, the similar to that of IEEE 802.11 for both ADMR and ODMRP.
node again remains stationary for the pause time, and then
chooses a new random location to proceed to at some new Packet Delivery Ratio
randomly chosen speed, and the node continues to repeat this 1
behavior throughout the simulation run. In [16] the authors 0,995
ADMR 1S1R
have proved that this model can produce large amounts of 0,99
Ratio

ADMR 1S3R
0,985
relative node movement and network topology change, and ODMRP 1S1R
0,98
ODMRP 1S3R
thus provides a good movement model with which to stress ad 0,975
hoc network routing protocols. 0,97
0 0,005 0,01 0,015 0,02 0,025
In the current ns-2 distribution, the speed is chosen Node Density (Node/m2)
uniformly randomly from [0,V_max], for every mobile node.
In this work two aspects of mobility have been investigated:
Fig. 9 PDR comparison with varying network density - 802.11

50
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

Packet Delivery Ratio C. Overhead

1 The overhead observed is the total overhead incurred at the
0,98 MAC layer and the routing layer. For the routing layer this
ADMR 1S1R
0,96
includes the overhead of ADMR and ODMRP for setting up
Ratio

ADMR 1S3R
0,94

0,92
ODMRP 1S1R and maintaining the multicast tree or forwarding group. As
ODMRP 1S3R
0,9 explained above all the control packets used by a specific
0,88 protocol are considered.
0 0,005 0,01 0,015 0,02 0,025
Network Density ( Node/m2)
As the simulation results prove ODMRP has an order higher
overhead mainly due to periodic flooding of join queries to
maintain redundant paths from source to destination. ADMR
Fig. 10 PDR comparison with varying network density -802.15.4
creates much lower overhead, independent of the network size
or the underlying MAC protocol. Another important
B. Varying Number of Senders and Receivers observation is that while for IEEE 802.11 networks ODMRP’s
The number of nodes in the network is set to 30 and the overhead is varying from 33% to 37% it is much higher for
nodes are static. The number of S is taken from the set {1, 3, 5, IEEE 802.15.4, reaching 53% (Fig.13, 14).
10, 15}. For MANET this is a model of “a class lecture
scenario”, while for WSNs (IEEE 802.15.4), a 1S represents “a
single node reading scenario”; 15S represent “a video Overhead

conference scenario” or “a single sink scenario” where 0,4

readings from 15 nodes are sent to a single sink node. 0,35

0,3
ADMR 1 R
Respectively the case with several receivers represents “a 0,25
Ratio

ADMR 5 R
0,2
multi-sink scenario”. 0,15
ODMRP 1 R
ODMRP 5 R
It is observed that the performance is much more stable for 0,1
0,05
both network protocols under IEEE 802.11. For wireless 0
0 10 20 30 40 50 60
sensor networks ODMRP has a varying behavior. It is claimed Number of Nodes
in [5] that ODMRP performs well in MANETs for greater
number of receivers and this is in line with our observations. Fig. 13 Overhead as a function of network size–802.15.4
Unfortunately the same cannot be claimed for WSN. The PDR
in the latter is reduced by nearly 10% compared to that in Overhead

MANETs. ADMR shows a much more consistent performance 0,6

for both MAC layer protocols. (Fig.11 and Fig.12). 0,5

0,4 ADMR 1 R
Ratio

ADMR 5 R
0,3
ODMRP 1 R
0,2 ODMRP 5 R
0,1
Packet Delivery Ratio- 30 Node 1 Receiver Varying Senders
0
1 0 10 20 30 40 50 60
0,995 Number of Nodes
0,99 ADMR 802.15.4
Fig. 14 Overhead as a function of network size–802.11
Ratio

0,985 ADMR 802.11

0,98 ODMRP 802.15.4
0,975 ODMRP 802.11

0,97
0,965
D. Impact of Mobility
0 5 10 15 20
Number of Senders
For studying the impact of mobility the network size is
constant at 30 nodes, the node mobility speed is varied 2, 10,
Fig. 11 PDR for a varying number of senders and 15 m/s, and pause time is 0. The impact of mobility is
evaluated by means of PDR and OR metrics. To create a
suitable model of a sensor network, 1 receiver and a varying
Packet Delivery Ratio- 30 Node 1 Sender Varying Receivers number of senders (1, 5 and 15) is selected.
1 The achieved results (Fig.15 and Fig.16) support the ones in
0,99 [5] that ODMRP is more efficient in more dynamic
ADMR 802.15.4
0,98 environments. This is more evident in WSNs. The PDR
Ratio

ADMR 802.11
ODMRP 802.15.4
0,97
ODMRP 802.11
achieved using ODMRP is around 93% for 5s at 15m/s while
0,96
that with ADMR is only around 83%. On the other hand,
0,95
0 5 10 15 20 greatly increasing the number of senders (15) together with
Number of Receivers their speed reduces the PDR noticeably for both protocols.
Fig. 12 PDR for a varying number of receivers

51
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

Packet Delivery Ratio - 5 and 15 senders VI. CONCLUSION

In this paper we have provided a comparative performance
0,999
0,997 study of two multicast protocols, ADMR and ODMRP, over
ADMR 5s
0,995
two different underlying MAC layer protocols – the IEEE
Ratio

0,993 ADMR 15s

0,991 ODMRP 5s 802.11 and the IEEE 802.15.4. The impact of node density,
0,989 ODMRP 15s

0,987
changing number of senders and receivers and mobility speed
0,985 on the PDR has been studied. One of the important
0 2 4 6 8 10 12 14 16
Node Speed
conclusions is that while these two routing protocols show
quite a stable performance for different scenarios based on
Fig. 15 PDR for a varying number of senders at different speeds – IEEE 802.11 the same is not true for the case of IEEE
802.11
602.15.4. Even though their operation is independent of the
underlying MAC layer, it is observed that both the PDR and
Packet Delivery Ratio - 5 and 15 senders the overhead values are quite sensitive to the media access
1 control. This study points out to some specifics when utilizing
0,95
higher level protocols designed for Ad Hoc networks in WSN.
ADMR 5s
It also supports the thesis that there is a strong relation
Ratio

ADMR 15s
0,9
ODMRP 5s between the contention mechanism used for media access and
ODMRP 15s
0,85 the performance of the routing protocol both in terms of packet
0,8 delivery ratio and overhead, with either static or mobile sensor
0 2 4 6 8 10 12 14 16
Node Speed
nodes.

Fig. 16 PDR for a varying number of senders at different speeds – REFERENCES

802.15.4
[1] K. Murray, A. Timm-Giel, M. Becker, C. Guo, R. Sokullu, D. Marandin,
“The European Network of Excellence CRUISE Application Framework
The total incurred overhead (Fig.17,18) for both IEEE and Network Architecture for Wireless Sensor Networks,” in the Proc. of
802.11 and 802.15.4 is little influenced by increasing the node WS on WMSN, GLOBECOM 2007, UAS, Nov. 2007.
[2] IEEE 802.11, Part 11: Wireless LAN Medium Access Control (MAC)
speed. But for ODMRP there is 35% to 40% overhead in Ad and Physical Layer (PHY) specifications, IEEE Std. 802.11-1999, Aug.
Hoc networks while in WSN it is as high as 70%. There is also 1999.
difference whether we have a large number of senders or a [3] S Kim, SJ Lee, S Choi - Wireless Mesh Networks, 2006. WiMesh 2006.
25 September 2006, Reston, Virginia,
large number of receivers. For 15s-1r at 15 m/s the overhead in [4] Wireless medium access control and physical layer specifications for low
ODMRP is round 70% compared to only 45% for 1s-15r at rate wireless personal area networks. Technical report, Part 15.4,
15m/s. ANSI/IEEE Standard 802.15.4, Sept. 2003.
[5] S. Coleri Ergen, ZigBee/IEEE 802.15.4 Summary, September 2004.
[6] Intra- and InterDomain Routing in the NGI, Euro-NGI, 507613, Sixth
Overhead- 30 Nodes, 5 and 15 senders, 5 and 15 Framework programme, Deliverable reference number: D.WP.JRA.1.3.3
receivers
[7] Y. Zhu, Pro-Active Connection Maintenance in AODV and MAODV,
0,4 MSc Thesis in Information and Systems Science, Carleton University,
0,35
0,3
ADMR 5s Ottawa, 2002.
ADMR 15s
0,25 [8] Kumar Viswanath, Katia Obraczka and Gene Tsudik,Exploring Mesh-
Ratio

ODMRP5s
0,2
ODMRP 15s and Tree Based Multicast Routing Protocols for MANETs , Transaction
0,15
0,1
ODMRP 5r of Mobile Computing (TMC 2004)
ODMRP 15r
0,05 [9] S.-J. Lee, M. Gerla, and C.-C. Chiang, “On-demand multicast routing
0
0 5 10 15 20
protocol,” in the Proceedings of the Wireless Communications and
Node Speed Networking Conference, WCNC, New Orleans, September 21-24, 1999.
pp. 1298-1302 and IETF Internet draft (expired), draft-ietf-manet-odmrp-
Fig. 17 Overhead for a varying number of senders and receivers at 04.txt, November 2002.
different speeds – 802.11 [10] J. G. Jetcheva and D. B. Johnson, “Adaptive Demand-Driven Multicast
Routing in Multi-Hop Wireless Ad Hoc Networks,” in the Proceedings of
the ACM Symposium on Mobile Ad Hoc Networking and Computing
Overhead - 30 nodes, 5 qand 15 sendersl 5 and 15
(MobiHoc), Long Beach, CA, October 2001.
receivers
[11] J. G. Jetcheva and D. B. Johnson, “The Adaptive Demand-Driven
0.8
ADMR 5s
Multicast Protocol for Mobile Ad Hoc Networks,” IETF Internet Draft
(expired), draft-ietf-manet-admr-00.txt, July 2001.
Overhead Ratio

0.6 ADMR 15s

ODMRP5s [12] Network Simulator - ns-2, http://www.isi.edu/nsnam/ns/
0.4
ODMRP 15s [13] S.-J. Lee, W. Su, J. Hsu, M. Gerla, and R. Bagrodia, “A Performance
0.2 ODMRP 5r
Comparison Study of Ad Hoc Wireless Multicast Protocols,” in the
ODMRP 15r
0 Proceedings of the Nineteenth Annual Joint Conference of the IEEE
0 5 10 15 20
Computer and Communications Societies, INFOCOM 2000, Tel Aviv,
Node Speed
pp. 565-574, March 2000.
Fig. 18 Overhead for a varying number of senders and receivers at [14] J.Zheng, M.J.Lee, Comprehensive performance study of 802.15.4,
Technical Report, [email protected], [email protected]
different speeds – 802.15.4 [15] J. Broch, D.A. Maltz, D.B. Johnson, Y.C. Hu, and J. Jetcheva, “A
Performance Comparison of Multi-Hop Wireless Ad Hoc Network

52
 INTERNATIONAL JOURNAL OF COMMUNICATIONS
Issue 1, Volume 2, 2008

Routing Protocols,” In Proceedings of the Fourth Annual ACM/IEEE

International Conference on Mobile Computing and Networking
(MobiCom’98), October 25–30, 1998, Dallas, Texas, USA.
[16] Atanasov I., E. Pencheva, “A mark-up approach to service creation”,
African Journal on Information and Communication Technology, ISSN
1449-2679, vol.3 No 1, March 2007, pp 8-18
[17] R. Sokullu, E. Karatepe, “Adaptive Packet selection algorithms for
Bluetooth Data Packets”, Proc. of the 6th WSEAS International
Conference on Applied Computer Science, Hangzhou, 15-17 April, 2007,
pp. 160-164
[18] D.B. Johnson, D.A. Maltz, J. Broch, “DSR: The Dynamic Source
Routing Protocol for Multi-Hop Wireless Ad Hoc Networks,” Mobile
Computing, 10.1007/978-0-585-29603-6_5, pp. 153-181.
[19] T-S. Chen, C-C Lin, Y-H Chiu, “A New Parameter-Free Classification
Algorithm Based on Nearest Neighbor Rule and K-means for Mobile
Devices”, Proc. of the 6th WSEAS International Conference on Applied
Computer Science, Hangzhou, 15-17 April, 2007, pp. 151-154
[20] E. Berg, S. Madhani, T. Zhang, Real-Time Path Quality Comparison for
QoS-Aware Roaming Between Heterogeneous Radio Networks, WSEAS
Transactions on Systems, ISSN1109-2777, Issue 4, Volume 3, June 2004,
pp. 1796 -1800

Radosveta Sokullu (M’85–PhD’91) has received an MSc in Radiotechnology

and a PhD degree in Telecommunications from the Technical University Sofia.
She is a member of NAUN since 2008. Currently she is with the Department of
EEE at Ege University, holding the position of Head of the Chair of
Telecommunications. Her major research interests are in the field of wireless
communications, cross layer design and wireless sensor networks.

Ozlem Karaca (M’04) has received an MSc degree in Telecommunications

from the Ege University. She is PhD candidate at Ege University. Currently
she is working in Telecommunication branch at Izmir Vocational School,
Dokuz Eylul University. Her major research interests are in the field of
wireless communications, routing algorithms, cross layer design and wireless
sensor networks.

53