Dead Node Detection using Particle Swarm

Optimization

Abstract---In computer science, particle treated as a point in a N-dimensional space

swarm optimization (PSO) is a which adjusts its “flying” according to its
computational method that optimizes a own flying experience as well as the flying
problem by iteratively trying to improve a
experience of other particles. It Combines
candidate solution with regard to a given
measure of quality. In PSO, simple self-experiences with social experiences.
software agents, called particles, move in
the search space of an optimization 2. BASICS OF PSO
problem. This paper deals with the dead
node detection and finding an optimized PSO is a Collection of flying particles
path without dead nodes. (swarm) where each particle keeps track of
its best solution, personal best, pbest and the
Keywords- Swarm, Dead nodes, PSO, best value of any particle, global best, gbest.
network, Energy parameter, Distance, Each particle adjusts its travelling speed
Iteration, Inertial weight factor, PSO
dynamically corresponding to the flying
algorithm.
experiences of itself and its colleagues. In
brief,

•
1. INTRODUCTION Each particle keeps track of its
coordinates in the solution space
Particle swarm optimization (PSO) is a which are associated with the best
population-based stochastic approach for solution (fitness) that has achieved
solving continuous and discrete optimization so far by that particle. This value is
problems. PSO is a robust stochastic called personal best, pbest.
optimization technique based on the •
Another best value that is tracked by
movement and intelligence of swarms. PSO the PSO is the best value obtained so
applies the concept of social interaction to far by any particle in the
problem solving. It was developed in 1995 neighborhood of that particle. This
by James Kennedy (social-psychologist) and value is called gbest.
Russell Eberhart (electrical engineer). It uses
•
a number of agents (particles) that constitute The basic concept of PSO lies in
a swarm moving around in the search space accelerating each particle toward its
looking for the best solution. Each particle is pbest and the gbest locations, with a
 random weighted accelaration at satisfactory solution will eventually be
each time step. discovered.

2.1 Comments on the inertial weight Formally, let f: ℝn → ℝ be the cost function
factor: which must be minimized. The function
A large inertia weight (w) facilitates a takes a candidate solution as argument in the
global search while a small inertia weight form of a vector of real numbers and
facilitates a local search.By linearly produces a real number as output which
decreasing the inertia weight from a indicates the objective function value of the
relatively large value to a small value given candidate solution. The gradient of f is
through the course of the PSO run gives the not known. The goal is to find a solution a
best PSO performance compared with fixed for which f(a) ≤ f(b) for all b in the search-
inertia weight settings. space, which would mean a is the global
Larger w means greater global search ability minimum. Maximization can be performed
and smaller w shows greater local search by considering the function h = -f instead.
ability.
w=wMax-[(wMax-wMin)xiter]/maxIter(2) Let S be the number of particles in the
where, swarm, each having a position xi ∈ ℝn in the
search-space and a velocity vi ∈ ℝn. Let pi be
 wMax= initial weight, the best known position of particle i and let
 wMin = final weight, g be the best known position of the entire
swarm. A basic PSO algorithm is then:
 maxIter = maximum iteration
number,  For each particle i = 1, ..., S do:
o Initialize the particle's
 iter = current iteration number. position with a uniformly
distributed random vector:
xi ~ U(blo, bup), where blo and
bup are the lower and upper
boundaries of the search-
3. PSO ALGORITHM space.

A basic variant of the PSO algorithm works o Initialize the particle's best
by having a population (called a swarm) of known position to its initial
candidate solutions (called particles). These position: pi ← xi
particles are moved around in the search-
If (f(pi) < f(g)) update the
space according to a few simple formulae. o

swarm's best known position:

The movements of the particles are guided
g ← pi
by their own best known position in the
search-space as well as the entire swarm's o Initialize the particle's
best known position. When improved velocity: vi ~ U(-|bup-blo|, |bup-
positions are being discovered these will blo|)
then come to guide the movements of the
swarm. The process is repeated and by doing  Until a termination criterion is met
so it is hoped, but not guaranteed, that a (e.g. number of iterations performed,
 or a solution with adequate objective traffic increases over the network such type
function value is found), repeat: of network suffers from the problems like
congestion and packet loss. Over the period
o For each particle i = 1, ..., S these sensors loss the energy and can be
do: dead. If the network is wireless in such case
there are more chances of inclusion of some
 For each dimension
external node in the network. In such case
d = 1, ..., n do:
there are more chances of Intrusion or
 Pick random packet loss over the network. The packet
numbers: rp, rg loss is acceptable up to some threshold value
~ U(0,1) but as there is more packet loss we need
some solution for this. The same solution is
 Update the presented in this proposed work. The
particle's complete solution is defined in terms of two
velocity: stages. In first stage we have to find the
vi,d ← ω vi,d + node that is responsible for packet loss over
φp rp (pi,d-xi,d) the network. Another stage is the
+ φg rg (gd-xi,d) development of algorithm or approach that
will eliminate the node dynamically and get
 Update the particle's the reliable data transmission over the
position: xi ← xi + vi network. As we are working on broken link
If (f(xi) < f(pi)) do: problem in a peer to peer network. Because

of this we have to first establish a sensor
 Update the network. For implementing this work we
particle's best have used MATLAB editor tool. The result
known
position:
pi ← xi

 If (f(pi) < f(g))

update the
swarm's best
known
position:
g ← pi
 Now g holds the best found solution.

The parameters ω, φp, and φg are selected by

the practitioner and control the behavior and
efficacy of the PSO method.

4. DEAD NODE DETECTION

A Wireless Sensor network is a dynamic

network with large no. of nodes. As the
will be shown in command window and in performed. With each transmission some
figures. energy gets lost associated with each node.
As the energy of a node get 0, the node is set
Figure 4.1 Network establishment as dead node.

We had taken 15 nodes. The above figure Figure 4.3 Network with dead nodes
shows their X and Y axis values assigned to
In the above figure, the red dots denote the
them. All of these nodes have some energy
dead nodes which are also shown with green
values which also shown in the figure.
color cloud callout. Here two nodes are
Based on these parameter values, dead node
shown as dead nodes. It means after the
detection will be done.
communications performed, the energy of
these two nodes got zero. So these are
declared as the dead nodes.

Figure 4.2 Network established having 15

nodes
The above figure has shown the network
established which consists of 15 nodes. Now
we apply the dead node detection algorithm
on this network. According to it, a number
of iteration are performed. Here the number
of iteration means total number of
communication performed. It includes the
number of time data transmission is Figure 4.5 Communication path between
performed. We can take any value for it. 15 nodes
Here we had taken its value 100 means
hundred times communication will be The above figure shows the communication
path between the 15 nodes in the network,
when there no optimization is performed. It
includes both the live and dead nodes. All of
fifteen nodes are shown separately with
naming given as node1, node2…and so on.
In this communication path, maximum
distance is covered while traversing the
nodes. As dead nodes are also included so
whenever this path will be followed for Figure 5.2 Without dead nodes
communication, it will consume a lot of time
and energy. Whenever we reach to dead Figure 5.1 and 5.2 show the graph between
node, the communication will be stopped. distance covered during each
There will be no use of finding this kind of communication is performed. The X axis
communication path including dead nodes. shows the no. of cycles or iterations or no.
of communication is performed. While the Y
5. ANALYSIS OF IMPLEMENTED axis shows the distance covered with respect
WORK to each iteration. Figure 5.1 stands for
network with dead and live both nodes. As
We will do analyses of our work with shown in it, distance covered is large.
respect to the distance covered with and Because no optimization is performed, due
without dead nodes. Both of these cases will to it there will be only wastage of time and
be understand with help of graph shown energy. Figure 5.2 stands for the network
below. with dead node detection and after
discarding these dead nodes. It clearly
shows that the distance covered is much less
than the previous figure. Optimization is
performed here. The cost factor used for
optimization is distance covered. In this way
swarm optimization is performed for dead
node detection and then discarding of these
dead nodes.
6. FUTURE WORK
In this work, we had generated values of X
and Y axis for nodes randomly. Also the
energy given to each node is randomly
generated with each new running task of the
work. In this work we can also change the
Figure 5.1 With dead nodes no. of nodes taken and no. of iteration
performed. For the future work, we can
change the optimization factor and then
perform this optimization again so that with
the distance covered in communication path,
also the time taken for the optimized path
also become optimum.

References
[1] Kennedy, J. “The particle swarm: Social
adaptation of knowledge”, Proceedings of
the 1997 International Conference on
Evolutionary Computation, IEEE Service
Center, Piscataway, NJ. Pp.303-308(1997).
[2] Brandstatter, B. and Baumgartner, U.,
"Particle swarm optimization-mass-spring
 system analogon," IEEE Transactions on consumption by the particle swarm
Magnetics, vol. 38, no. 2, pp. 997-1000, optimization,” in Proceedings of the IEEE
(Mar.2002). Swarm Intelligence Symposium (SIS), pp.
[3] El-Gallad, El-Hawary, Sallam, and Kalas, 23–28,(2007).
“A Particle swarm optimizer for constrained [11]A. Boukerche, H. A. B. Oliveira, E. F.
economic dispatch with prohibited operating Nakamura, and A. A. F. Loureiro,
zones”, Canadian Conference on Electrical “Localization systems for wireless sensor
and Computer Engineering, 2002, pp. 78- networks,” IEEE Wireless Commun. Mag.,
81(2002). vol. 14, no. 6, pp. 6–12, (December 2007).
[4] Hu, X. and Eberhart, R. C., “Multiobjective
optimization using dynamic neighborhood
particle swarm optimization”,. Proceedings
of the IEEE Congress on Evolutionary
Computation (CEC 2002), Honolulu,
Hawaii USA. pp. 1677-1681, (2002)
[5] Cedeno, W. and Agrafiotis, D. K., "Using
particle swarms for the development of
QSAR models based on K-nearest neighbor
and kernel regression," Journal of
Computer-Aided Molecular Design, vol. 17,
no. 2-4, pp. 255-263, (Feb.2003).
[6] Mostaghim, S. and Teich, J. r. “Strategies for
finding local guides in multi-objective
particle swarm optimization (MOPSO)”,
Proceedings of the IEEE Swarm Intelligence
Symposium 2003 (SIS 2003), Indianapolis,
Indiana, USA. pp. 26-33, (2003).
[7] Zheng, Y., Ma, L., Zhang, L., and Qian, J.
“Empirical study of particle swarm
optimizer with an increasing inertia weight”,
Proceedings of IEEE Congress on
Evolutionary Computation 2003 (CEC
2003), Canbella, Australia. pp. 221-226,
(2003).
[8] Zhang, X., Yu, L., Zheng, Y., Shen, Y., Zhou,
G., Chen, L., Xi, L., Yuan, T., Zhang, J., and
Yang, B., "Two-stage adaptive PMD
compensation in a 10 Gbit/s optical
communication system using particle swarm
optimization algorithm," Optics
Communications, vol. 231, no. 1-6, pp. 233-
242, (Jan.2004).
[9] X. Wang, S. Wang, and J. J. Ma, “An
improved co-evolutionary particle swarm
optimization for wireless sensor networks
with dynamic deployment,” Sensors, vol. 7,
pp. 354–370, (2007).
[10] T. P. Hong and G. N. Shiu, “Allocating
multiple base stations under general power