You are on page 1of 4

Proceedings of 2015 Global Conference on Communication Technologies(GCCT 2015)

Decentralized Topology Management on Mobile Ad


hoc Networks
Aniket Pramanik1 , Biplav Choudhury2 , Tameem S. Choudhury3 , Wasim Arif4 , J. Mehedi5
Department of Electronics and Communication Engineering
National Institute of Technology, Silchar
Silchar, Assam-788010, India
aniketpramanik@yahoo.co.in1 , biplav93@gmail.com2 , salmantameem360@gmail.com3 , arif.ece.nits@gmail.com4 ,
j.mehedi@gmail.com5
AbstractIn this paper, a new simulation tool is described
which is used to present a decentralized approach on Mobile
Ad Hoc Networks. The simulation is developed in Matlab using
a decentralizing algorithm. The paper includes a distributed
algorithm for adaptive movement of nodes in a MANET (Mobile
Ad hoc Network) to maintain the overall topology of the network.
The proposed algorithm assumes the presence of a GPS receiver
in each node so that every node is aware of the position of
every other node. A node measures the relative position and
velocity of its neighbours and checks whether a certain critical
condition is satised or not. If the condition is satised, it keeps
its position unchanged; otherwise a rush call is made to that
node by its nearest neighbour to get it back within the range.
The overall advantage here is that our new tool doesnt involve
any co-ordinator which signicantly reduces the overhead on the
network. The simulation of the algorithm has been carried out on
a few synthetically generated network scenarios and the results
thus obtained show the effectiveness of the proposed algorithm.
Keywords: MANET, Topology Management, Distributed Algorithm

I.

I NTRODUCTION

Mobile Ad-hoc Networks (MANETs) are wireless networks consisting of mobile nodes that communicate on-themove without base stations [1]. Nodes in these networks
generate both user and application trafc and carry out network
control and routing protocols. Rapidly changing connectivity,
network partitions, higher error rates, collision, interference,
bandwidth and power constraints together pose new problems
in network control, particularly in the design of higher level
protocols such as routing and in implementing applications
with Quality of Service requirements [2]. The simulation
tool, Decentralized Hop Propagation Coordinator Independent
Algorithm provides node movement in a decentralized manner.
There is no need of any coordinator to call the nodes which are
going out of range. This paper is started with the description of
brief background of an ad hoc network and the motivation for
using decentralized approach. After that, the whole algorithm
is described, which involves calling of nodes without coordinator. Finally, the stimulation results conrm the validity and
proof of the algorithm in the network.
II.

radios collaborate to create their own network. When a user


turns on his radio, that radio follows a protocol to announce
its own existence and to discover other radios. This approach
to networking is necessary for tactical networks that cannot
rely on infrastructure, such as those used by military and
disaster relief organizations [2], [5], [6]. It is also appropriate
for sensor networks and inter-vehicular networks. MANETs
present many challenges to networking protocols. In the absence of centralised control by base stations and access points,
protocols must provide distributed solutions. The mobility of
nodes causes those solutions to be eeting, therefore inter-radio
communications for network maintenance persists throughout
the lifetime of the network. This is not more important than
routing protocols that track network topology. The plethora
of routing protocols proposed for MANET attests to the
challenge of this task. There is a direct correlation between
the effectiveness of a protocol to track topology and the
quantity of administrative trafc and this quantity increases
when either the size or volatility of the network increases.
However, wireless networks are capacity constrained.
III.

P ROBLEM D EFINITION

The aim is to provide node movement in a decentralized


manner, where there is no involvement of a leading coordinator. The network movement should begin from a place, where
all nodes are in position and are ready to go in a random
manner to the destination. Assumptions of the problem are
pointed below:

Four nodes are considered in the program keeping in


focus that the same algorithm is valid for more nodes
too.

There is no coordinator considered in the algorithm.


Each node has the ability to communicate to its
neighbours. This makes the network decentralized.

It is also assumed that the nodes present in the


network move in random directions but as a whole
they move towards a destination.

The calling mechanism used in the algorithm is such


that, the nodes get a call to come to range whenever
they are out of the range of their neighbours. In the

BACKGROUND

Mobile ad hoc networks (MANETs) are wireless networks


without infrastructure [3], [4]. Rather than connecting to a
base station, as in wireless telephony or an access point, user

978-1-4799-8553-1/15/$31.00 2015 IEEE



Proceedings of 2015 Global Conference on Communication Technologies(GCCT 2015)

Fig. 1.

Node positions for case 1

Fig. 2.

program, a centroid is considered to be the point


where a node is called which is going out of range.
This makes the node communicable to other nodes
present in the network.

IV.

Centroid is calculated by taking individual mean of


X Coordinates and Y Coordinates (As the network is
not geometrical). Here as shown in Fig. 1 the one with
X Mark is the centroid while others are nodes in the
network.
A LGORITHM I MPLEMENTATION AND E XPLANATION

Given below are the steps followed by the network to


move towards the destination. Considering all the introduced
assumptions we follow the steps given below:

Initially, four nodes are randomly placed at 10


percent closer to X coordinate, and Y coordinate
has a maximum value of 1500 from which random
coordinate value is allocated for each node in Y. This
provides the initial random distribution of nodes. The
network distribution comes out to be as Fig.2 where
X mark in the gure is the centroid of the network.
After going through the gure it is found that either
pairwise Node A and Node C are out of range or
node pair, Node B and Node D are out of range . So
if communication exists between Node A and Node
D (look at Red and Yellow in the Fig. 2), either Node
A is called by Node D or vice-versa to the centroid
marked by X. In case of more nodes, majority nodes

978-1-4799-8553-1/15/$31.00 2015 IEEE

Node positions for case 2

will call up other nodes which are out of range. The


caller sends a Rush Message to the receiver. Assume
that Node A is called; then the distance between
Node A (Red) and Node C (Navy Blue) increases
which makes Node D out of range. So the node that
is closer to Node D sends a Rush Message to Node
D, and here it is Node A which sends the message.

After the rst step the communication between all four


nodes is possible. Now, the nodes start moving in a
random fashion. The random movement is followed by
a probability function that is introduced in the program
in Matlab. It is assumed that the whole network moves
towards Y direction so there should be increase in
movement of nodes mostly towards the Y rather than
X. Movement of X is followed by
X = X + Random(15; 15)

(1)

This random function obeys uniform distribution in


the program. So, it gives them freedom to move in X
direction. Similarly for Y direction,
Y = Y + Random(4; 4)

(2)

This function makes the nodes move in upward


Y - direction. This random function is unique for
each node. Hence the movement algorithm is totally
randomized.



Proceedings of 2015 Global Conference on Communication Technologies(GCCT 2015)

Fig. 3.

Fig. 5.

Distance vs Critical Distance for Node B

Fig. 6.

Distance vs Critical Distance for Node C

Node positions for case 3

V.

Fig. 4.

Distance vs Critical Distance for Node A

After these nodes start moving with different


velocities, there may be possibility that these get out
of range. As shown, the Node C (Navy Blue) gets
out of range. The nearest to it is Node B which sends
a message to Node C to get in range.

In this way, the nodes continue their movement


towards the destination without getting diverted from
the network. Similarly, if the same algorithm is
considered for n number of nodes, the result will be
identical.

978-1-4799-8553-1/15/$31.00 2015 IEEE

S IMULATIONS AND R ESULTS

The four different plots obtained as a result of the simulation are shown in Fig. 4,Fig. 5, Fig. 6 and Fig. 7 where
each graph is plotted as a distance of other nodes from
itself. For example in the rst plot, Fig. 4, distance of other
nodes from node A as measured during the simulation are
plotted. In the gure, the Dark Blue line represents the distance
between node A and node D. Node D initially starts much
closer to node A at time=0 second, but at time=500 seconds
(approximately), it crosses the critical distance (marked by a
Sky Blue line) which is considered as the maximum limit
to maintain the networks connectivity. So, it gets a call
from Node B as shown in the calling result on Table 1, to
come within the neighbourhood following which the distance
between node A and D decreases as evident from the Fig.
4. Similarly, Fig. 5 is for Node B. Here too, when the node
D, represented by Dark Blue, crosses the critical distance at
time=400 seconds(approximately), node A calls it and node D
immediately comes within the neighbourhood. It is followed



Proceedings of 2015 Global Conference on Communication Technologies(GCCT 2015)

number of nodes surrounding a single node. The movement of


nodes implemented, follows Brownian motion and the whole
movement was plotted. It validates the algorithm on node
movement.
R EFERENCES

Fig. 7.

Distance vs Critical Distance for Node D


TABLE I.

C ALLING R ESULT BASED ON S IMULATION


B calls D
B calls D
A calls D
D calls A
B calls D
D calls A
C calls B
A calls B
A calls B
C calls D
C calls A
B calls D
D calls B
B calls A
C calls D
A calls D
D calls A
A calls B
D calls B
C calls A
D calls B
B calls D
B calls D
B calls A
C calls B
A calls D
C calls A

A calls D
C calls A
D calls B
C calls B
B calls A
C calls D
D calls A
C calls B
C calls B
A calls D
B calls A
B calls D
A calls D
B calls A
B calls D
A calls B
C calls D
D calls A
C calls B
A calls D
A calls D
B calls A
B calls A
A calls B
D calls A
C calls B
C calls B

[1] C. S. R. Murthy and B. Manoj, Ad hoc wireless networks: Architectures


and protocols. Pearson education, 2004.
[2] S. Samanta, S. Ray, S. SenGupta, and M. Naskar, A novel algorithm
for managing network conguration, in Asian International Mobile
Computing Conference, 2006, pp. 0407.
[3] T. S. Rappaport et al., Wireless communications: principles and practice.
prentice hall PTR New Jersey, 1996, vol. 2.
[4] E. M. Royer and C.-K. Toh, A review of current routing protocols for ad
hoc mobile wireless networks, Personal Communications, IEEE, vol. 6,
no. 2, pp. 4655, 1999.
[5] K. Kim, Y. Cai, and W. Tavanapong, Sharing location dependent
experiences in manet, in Mobile Data Management, 2006. MDM 2006.
7th International Conference on. IEEE, 2006, pp. 6969.
[6] J. Liu, D. Sacchetti, F. Sailhan, and V. Issarny, Group management
for mobile ad hoc networks: design, implementation and experiment,
in Proceedings of the 6th international conference on Mobile data
management. ACM, 2005, pp. 192199.

by third and fourth plot for Node C and Node D respectively.


Whenever a node crosses the critical distance (assumed to
be 100 units throughout the simulation), it gets a call from
the node that is nearest to it. After getting the call, the node
crossing the critical distance gets within the neighbourhood
and in this way, the whole network maintains its connectivity
.The critical distance line is marked as a Sky Blue coloured
straight line (at x=100 units) on each of the plots. The calling
is done only when the node(s) are out of range, and these calls
are based on the same simulation result as shown.
VI.

C ONCLUSION

Going through other decentralized network approach algorithms, it is found that the calling mechanism is not appropriate
if the front node which goes out of the network is allowed to
stop and wait for other nodes to approach there. This problem
is totally eliminated in the proposed algorithm. Also, when the
number of nodes increases, the complexity depends only on the

978-1-4799-8553-1/15/$31.00 2015 IEEE



You might also like