emerging_2015_3_20_50041

Dynamic Node Movement Control in a Mobile Medium Ad hoc Network
Hanin Almutairi, John DeDourek, Przemyslaw Pochec
Faculty of Computer Science
University of New Brunswick
Fredericton, Canada
e-mail: {hanin.almutairi, dedourek, pochec}@unb.ca
Abstract—A Mobile Ad hoc Network (MANET) is a network of
wireless mobile devices capable of communicating with one
another without any reliance on a fixed infrastructure. A Mobile
Medium Network is a set of mobile forwarding nodes
functioning as relays for facilitating communication between the
users of this Mobile Medium. The performance of the Mobile
Medium depends on the Mobile Medium node density,
distribution and movement. In the proposed dynamic node
movement, the movement is determined based on whether the
node is on a forwarding path for a data flow or not. Simulation
results show that slowing down the speed of mobile nodes when
they are forwarding significantly affects the delivery rates in
Mobile Medium networks. For networks with a few forwarding
nodes dispersed in a large region reducing the mode movement
speed by 50% results in an approximately 20% improvement in
the delivery ratio, with even higher improvements possible at
lower speeds.
Keywords-mobility models; Mobile Medium; self-organizing
mobile network; M2ANET; SMMANET; MANET; AODV;
DSDV; ns2 simulation
I. INTRODUCTION
A Mobile Ad hoc Network (MANET) is a set of mobile
devices that cooperate with each other by exchanging
messages and forwarding data [1][2]. A Mobile Medium Ad
hoc Network (M2ANET) proposed in [3] is a particular
configuration of a typical MANET where all mobile nodes
are divided into two categories: (i) the forwarding only nodes
(shown in black in Figure 1) forming the so called Mobile
Medium, and (ii) the communicating nodes (shown in red in
Figure 1), mobile or otherwise, that send data and use this
Mobile Medium for communication. The advantage of this
M2ANET model is that the performance of such a network is
based on how well the Mobile Medium can carry the
messages between the communicating nodes and not based
on whether all mobile nodes form a fully connected network.
An example of a M2ANET is a cloud of autonomous drones
released over an area of interest facilitating communication
in this area. The movement of nodes in a M2ANET can be
predefined by the user, selected at random or purposefully
controlled for the best performance. When the mobile nodes
select themselves their movement we refer to such a network
as a Self-organizing Mobile Medium Ad hoc Network
(SMMANET). Recently, a number of projects that match the
M2ANET model have been announced; they include Google
Loon stratospheric balloons [4] and Facebook high altitude
solar powered planes [5] for providing Internet services to
remote areas, and the Swarming Micro Air Vehicle Network
(SMAVNET) project where remote controlled planes are
used for create an emergency network [6].
Figure 1. ns2 simulation screen of a M2ANET
Controlling the movement of the forwarding nodes
forming the Mobile Medium is a problem in deploying
M2ANETs. Random movement, while easy to implement,
suffers from the difficulty of how to keep nodes in a sufficient
density over the area of interest where the communication
infrastructure is to be supplied by means of a Mobile
Medium. A Mobile Medium with low mobile node density
suffers from frequent disconnections and rerouting, resulting
in a low delivery ratio. In this paper, we propose a solution
for controlling the movement of the Mobile Medium nodes
for M2ANET (SMMANET) deployments. The solution is
based on the following observation: In a typical M2ANET
only a few (if any) mobile nodes are actively forwarding the
network traffic for a given flow at any given time. The longer
the forwarding path is maintained the better the delivery ratio.
We therefore propose to slow down the movement of the
nodes that are actively forwarding data in the M2ANET. The
new approach is then compared against the standard Random
Way Point (RWP) movement.
In Section II, we present background on Mobile Medium
networks and mobility patterns. The new movement pattern
based on changing the speed of mobile nodes that happen to
be on the forwarding path for a flow is discussed in Section
50Copyright (c) IARIA, 2015. ISBN: 978-1-61208-422-0
EMERGING 2015 : The Seventh International Conference on Emerging Networks and Systems Intelligence

III. Simulation experiments of this movement under different
scenarios are in Section IV. Finally, we present the
experimental results in Section V, followed by the conclusion
and future work.
II. STATE OF THE ART
A MANET is comprised of interconnected mobile nodes,
which make use of wireless communication links for multi-
hop transmission of data. They offer distinct advantages over
infrastructure based networks and are versatile for some
particular applications and environments. There are no fixed
or prerequisite base stations or infrastructures; therefore, their
set up is not time consuming and can be done at any time and
in any place. MANETs exhibit a fault-resilient nature, given
that they are not operating a single point of failure and are
very flexible. The deletion and addition of new nodes,
forming new links are a normal part of operation of a
MANET [1][7][8]. A group of nodes can facilitate
communication between distant stations by forming a Mobile
Medium, as introduced in [3].
Many mobility models have been proposed for recreating
the real world application scenarios of MANETs. A mobility
model attempts to mimic the movement of real mobile nodes
that change speed and direction with time. There are two
main types of mobility models currently used in simulation
of MANETs [2][9]: trace and synthetic. A trace uses actual
node movements that have been observed in a real system. In
the absence of traces, synthetic mobility models can be used.
The synthetic models attempt to realistically mimic the
movements of mobile nodes in mobile networks [2]. The
categorization of synthetic models is based on interactions
between the nodes and the environment in a mobile network
[2]: we can distinguish between individual node movements
and group node movements. Based on specific mobility
characteristics these models can be further classified into four
categories: models with temporal dependency, models with
spatial dependency, models with geographic restriction, and
random models [2]. In the mobility model with temporal
dependency the movement of a mobile node is affected by its
movement history. A node’s current movement is affected by
past movement such as in the Gauss Markov Model and the
Smooth Random Mobility model [2]. In mobility models
with spatial dependency, the mobile nodes tend to travel into
a group and are interdependent one on another. The
movement of a node is affected by surrounding nodes in
group mobility such as in the Reference Point Group Model
[2]. One implementation of a group mobility model uses the
attraction/repulsion principle to maintain the connections
between the mobile nodes: when nodes get too close (e.g.,
signal strength is high) they select the next move with a
preference in the opposite direction to the incoming signal
and conversely, when the signal gets weak, they turn back
[10]. Another class is the mobility models with geographic
restriction. There, the mobile node movement is limited to
certain geographical areas, such as streets or freeways, as for
example in the Pathway Mobility Model and the Obstacle
Mobility Model [2].
In simulation, a random mobility is often used as a
reference case scenario, mostly because of the relative ease
of implementing it in a simulator. One of these popular
models is the Random Way Point (RWP) model available in
ns2 [11]. Nodes are moved in a piecewise linear fashion, with
each linear segment pointing to a randomly selected
destination and the node moving at a constant, but randomly
selected speed.
III. DYNAMIC MOVEMENT CONTROL IN MOBILE MEDIUM
Recall that the Mobile Medium is a particular type of a
mobile network that is used for providing communication
services to mobile or stationary users. It consists of a cloud
of mobile nodes whose sole function is to forward data. Users
of this Mobile Medium connect to it by establishing a link to
one of its nodes and then send data to other users connected
through this medium (Figure 2). Our interest here is to assure
the best delivery ratio for the data sent form one user to
another through the Mobile Medium.
Figure 2. Nodes on the forwarding path
We start with a common reference scenario where a
number of Mobile Medium nodes move in a restricted region
(a rectangle) in random directions according to the RWP
model. The two user stations (nodes) that wish to
communicate are placed in the same region, remain stationary
and send data one to another through the Mobile Medium
(Figure 2). Because the Mobile Medium nodes move in and
out of range of the communicating stations the connections
break and the routing path changes, affecting the delivery
ratio.
The Dynamic Movement Control in Mobile Medium is an
attempt aimed at reducing the (number of) changes in the
routing path during the data transfer between the users.
Normally, the nodes in the RWP model move in random
directions and at random speeds independent of whether they
are forwarding packets or not. The faster they move the
quicker they get out of range and the sooner the connection
path breaks. In the proposed Dynamic Movement Control for

the Mobile Medium, the nodes that forward data are slowed
down. Note that this decision to slow down does not require
any global information or location data; it is solely based on
the forwarding status of the node. The easiest way to
determine if the node is in the forwarding state is to check if
it had forwarded a data packet recently: for example, for the
scenario with a Constant Bit Rate (CBR) traffic source a node
would maintain its forwarding status, and move at a reduced
speed, for at least the time defined by the inter packet interval
of the CBR source. For other scenarios, like the exponential
traffic sources, multiples of the average inter packet interval
can be used for determining the forwarding status
maintenance period for a node. In the ns2 simulation we used
the actual path determined by the simulator to determine
which nodes are in the forwarding state.
The evaluation of the proposed Dynamic Movement
Control for Mobile Medium was performed in the ns2
simulator. The simulator is implemented in c++ and each
simulation is controlled by an Object Oriented Tool
Command Language (OTCL) script. Unfortunately, the
OTCL scripting language does not provide any access to the
routing information used in the simulation. To access the
routing information used by the Ad hoc On-Demand Distance
Vector (AODV) protocol we modified the
void forward (aodv_rt_entry *rt,
Packet *p, double delay);
function in aodv.cc, and to get routing for the Destination-
Sequenced Distance Vector (DSDV) the function
void forwardPacket (Packet *p);
in dsdv.cc was modified.
IV. SIMULATION EXPERIMENT
A set of simulation experiments was conducted to
evaluate the proposed Dynamic Movement Control for the
Mobile Medium networks. Each simulation of a network
consists of a different number of nodes roaming (RWP
movement, average speed 4 m/s) in a square 1000 x 1000
meters. The node transmission range is 250m. The link data
rate is 1 Mbps. Every packet has a size of 512 bytes. The
buffer size at each node is 50 packets. Data packets are
generated following a Constant Bit Rate process [11]. The
source and destination nodes are stationary and located at
coordinates (50, 400) and (950, 600). The summary of the
simulation parameters used in ns2 is shown in Table 1. In
each experiment, the designated source node transmits to one
designated destination node for 500 seconds. For each node
density and speed, the experiment is run three times and the
average delivery ratio is reported.
TABLE I. SIMULATION PARAMETERS
Parameters
Simulator NS-2.34
Channel Type Channel / Wireless Channel
Network Interface Type Phy/WirelessPhy
Mac Type Mac/802.11
Radio-Propagation Type Propagation/Two-ray ground
Interface Queue Type Queue/Drop Tail
Link Layer Type LL
Antenna Antenna/Omni Antenna
Maximum Packet in ifq 50
Area (n * n) 1000 x 1000m
Source node location (50, 400)
Destination node location (950, 600)
Source Type CBR over UDP
packetSize_ 512
interval_ 0.05
Simulation Time 500 s
Routing Protocol AODV and DSDV
A. AODV performance
Reducing the speed of the forwarding nodes running the
AODV routing protocol results in an improved delivery ratio
for all but the very low node density (5 nodes over
1000x1000m region) scenario, Fig 3. The networks with a
very few nodes in the Mobile Medium, i.e., with 5 and 10
nodes, as expected show very low delivery ratio. Under
normal operating conditions for the Mobile Medium, with 20
or more mobile forwarding nodes, the performance improves
gradually. Networks with a large number of nodes, more than
30 in a 1000x1000m area, as expected work well at any node
speed with delivery ration ranging from 75% to almost 100%.
The most interesting case is the network with a moderate
number of nodes, 20 nodes over the area 1000x1000m. This
scenario demonstrates best the benefits of the Dynamic
Movement Control for Mobile Medium networks. In this
scenario the delivery ration is very low 30% for nodes moving
at the original speeds, but the performance improves
significantly to 85% when the speed is reduced to 10% of the
original speed.
The benefits are the highest in the low to medium range of
node densities, 10 to 20 nodes moving in a 1000x1000m
region, as depicted in Figure 4.
B. DSDV performance
DSDV performance improvement is similar to AODV:
low improvements at low node densities (5 to 10 nodes) and a
gradual improvement with the reduction of the speed of the
forwarding nodes at higher node densities, as shown in
Figures 5 and 6. In our experiments, the overall delivery ratio
is lower for DSDV than for AODV. This is a common
situation for the DSDV protocol when used in the ad hoc
networks with a dynamically changing topology [12] and can
be attributed to the proactive nature of the DSDV protocol
where distributing the routing information to all the nodes and
detecting a valid route can take a considerable time.

Figure 3. AODV delivery ratio.
Figure 4. Delivery ratio for AODV at three speeds: no change, with 50%
reduction and zero.
Figure 5. DSDV delivery ratio
Figure 6. Delivery ratio for DSDV at three speeds: no change, with 50%
reduction and zero.
Figure 7. Time to detect the first path, for two different protocols.
In our experiments, no routes were detected in networks
with 5 nodes and it took approximately 80 seconds to detect
the first route in the networks with 20 or more nodes running
DSDV, which is considerably slower than the path detection
delay observed for the AODV protocol, see Figure 7.
V. CONCLUSION AND FUTURE WORK
In this paper, we proposed a new node movement control
paradigm for a self-organizing MANET network. The
approach is particularly attractive for M2ANETs where the
goal is to create a Mobile Medium out of mobile forwarding
nodes, and use this Mobile Medium to facilitate data
communication between other users. The new mobility
control mechanism is based on slowing down the nodes that
are on the forwarding path for a flow. The performance of the
M2ANET/SMMANET increased significantly with the
lowering of the speed of the nodes actively forwarding the
data, with a higher relative improvement for the networks
with moderate node densities. The improvements were
observed in the networks running two different routing
protocols AODV and DSDV.
Based on our results, we suggest further testing self-
organizing M2ANET/SMMANET networks using different
experimental scenarios. In particular while our experiments
involved two stationary communicating nodes it might be
more realistic for some application scenarios to model them
as mobile. Combining the attraction/repulsion mechanism
proposed in [10] with the new Dynamic Control may also
result in further improvement of the performance of a
SMMANET.
ACKNOWLEDGMENT
This work is sponsored and funded by the Ministry of
Higher Education of Saudi Arabia through the Saudi Arabian
Cultural Bureau in Canada.
REFERENCES
[1] S. Basagni, M. Conti, S. Giordano, and I. Stojmenovic (Eds.), Mobile
Ad Hoc Networking, New York, Wiley-IEEE Press, 2001.
[2] F. Bei and A. Helmy, A survey of mobility models in wireless Ad hoc
Networks, University of California, USA,2004.
[3] J. DeDourek and P. Pochec, “M2ANET: a Mobile Medium Ad Hoc
Network”, Wireless Sensor Networks: Theory and Practice, WSN 2011,
Paris, France, Feb. 2011, pp. 1-4.
N
1
N
1

[4] H. Hodson, "Google's Project Loon to float the internet on balloons",
New Scientist, October 2013.
[5] J. Brustein, "Facebook’s Flying Internet Service, Brought to You by
Drones", Bloomberg Businessweek, March 4, 2014.
[6] A. Jimenez Pacheco, et al., "Implementation of a Wireless Mesh Network
of Ultra Light MAVs with Dynamic Routing", IEEE GLOBECOM 2012,
3rd International IEEE Workshop on Wireless Networking & Control for
Unmanned Autonomous Vehicles 2012, Anaheim, California, USA, 2012.
[7] D. P. Agrawal and Q. A. Zeng, Introduction to Wireless and Mobile
Systems, Thomson Engineering, 2010, 1591-1596.
[8] S. K. Sarkar, T. G. Basavaraju, and C. Puttamadappa, Ad Hoc Mobile
Wireless Networks, Principles, Protocols, and Applications, Auerbach
Publications Taylor & Francis Group, 2007.
[9] N. Aschenbruck, E. G. Padilla, and P. Martini, "A survey on mobility
models for performance analysis in tactical mobile networks", Journal of
Telecommunications and Information Technology, vol. 2, 2008, pp. 54-61.
[10] N. Alsalmi, J. DeDourek, and P. Pochec, "Self-organizing Mobile
Medium Ad hoc Network", The Fourth International Conference on Mobile
Services, Resources, and Users MOBILITY 2014, July 20 - 24, 2014 - Paris,
France, pp. 38-42.
[11] H. Ekram and T. Issariyakul, Introduction to Network Simulator NS2,
Springer, 2009.
[12] S. Mohapatra and P. Kanungo, "Performance analysis of AODV, DSR,
OLSR and DSDV Routing Protocols using NS2 Simulator", Procedia
Engineering (International Conference on Communication Technology and
System Design 2011), v. 30, 2012, pp. 69-76.

emerging_2015_3_20_50041

More Related Content

What's hot (15)

Viewers also liked (8)

Similar to emerging_2015_3_20_50041 (20)

emerging_2015_3_20_50041