ISSN 2321 3361 © 2020 IJESC

Research ArticleVolume 10 Issue No.4

Secured Robustness Optimization Scheme with Multi-Population

Co-Evolution for Distributed Wireless Sensor Networks
V. Sowmitha1, V. Prakasham2, V. Gopinath3, R. Sathyapriya4
Assistant Professor1, 2, 3, 4
Department of CSE
K. S. R Institute for Engineering and Technology, Tiruchengode, TamilNadu, India

Abstract:
Wireless sensor networks (WSNs) have been the popular targets for cyber attacks in various topologies. One type of network topology
for WSNs, the distribution networks, can effectively withstand random attacks in which the nodes in the topology are randomly
selected as targets. In the distribution networks should be restored by reconfiguring switches automatically with smart grid
technologies. Be that as it may, it is delicate to malicious attacks in which the nodes with high node degrees are chosen as targets.
Thus, how to improve the robustness of the distribution networks against malicious attacks becomes a critical issue. To handle this
issue, this paper proposes a Robustness Optimization plot with multi-populace Co-advancement for distributed wireless sensor
networks (SROCKS) to improve the heartiness of the dissemination networks with smart grid technologies. We build distribution
networks according to the characteristics of WSNs in the real-world environment. Then, we apply our SROCKS with novel crossover
operator and mutation operator to optimize the robustness of the distribution networks constructed for WSNs. For a distribution
networks, our proposed algorithm keeps the initial degree of each node unchanged such that the optimized topology fast screening
algorithm based on hitting set enumeration. In light of a notable measurement for the robustness against malicious assaults, our test
results show that SROCKS generally pairs the power of beginning sans scale WSNs, and beats two existing algorithms by about 16%
when the network size is enormous.

Index Terms: Wireless sensor networks, robustness, Distribution network, hitting set, multi-population co-evolution, restoration.

I. INTRODUCTION the most proficient method to improve scale-free topologies o

withstand malicious attacks. Some proposed approaches attempt
WIRELESS sensor networks (WSNs) [1] have become a hot
to improve the robustness of systems with Genetic Algorithm
research field with a broad range of applications. Commonly,
(GA) [5]. Due to single population of candidate solutions, it
WSNs send more number of network nodes inside a specific
brings a typical limitation called premature convergence [6], in
zone, and these nodes speak with one another to screen or
which the evolution falls into a local optimum too early,
control natural parameters, for example, temperature, lighting,
resulting in a solution far from the global optimum. Besides, it is
and so forth. Because of the pervasiveness of cyber attacks, how
known that multi-population genetic algorithm can effectively
to improve the robustness of WSNs becomes an essential issue
overcome this limitation by using multiple populations to
in recent years [1]. Network topology abstracts how nodes are
evolve together. Different probability of crossover operator and
connected in a network. It is the foundation for communication
mutation operator are assigned to each population of multiple-
activities happening inside the network. Complex network
population. Individuals with high fitness values can be
theory [2] studies the network topologies of an important class
introduced into other different populations through migration
of net- works called Complex Networks in which the topological
operator, which can effectively prevent falling into a local
feature is neither purely regular (e.g., in a lattice graph [2]) nor
optimum. Therefore, in this paper, we propose to use multi-
purely random (e.g., in a random graph [2]). Actually, many
population co-evolution to enhance the robustness of scale-free
kinds of real-world networks such as the Internet, social
topologies. To approve this idea, we give a solid plan called
networks, brain networks, etc. are all complex networks. There
Robustness Optimization with multi-population Co-development
are two classic models in complex network theory, one is small
for scale-free. This paper proposes a novel method that
world topology and the other is scale-free topology. The little
efficiently finds (nearly) all unrestorable cutsets. The proposed
world model has two remarkable features [3] which are little
method avoids the computation issues by employing a
average way length and high clustering coefficient [3]. It is
compressed data structure named the zero-suppressed binary
generally used in modeling heterogeneous network topologies in
decision diagram, or zero-suppressed binary decision diagram
WSNs [3]. Then again, scale free model is portrayed by the
(ZDD) [6], which allows us to represent a huge non convex
power law circulation of node degrees, and basically utilized in
space in a compressed manner, and also to execute efficient
displaying homogeneous network topologies [4]. Since the
algebra on the compressed space. ZDDs have been successfully
scale-free model has a force law appropriation of node degrees,
applied to the loss minimization problem of distribution
it is hearty against arbitrary attacks, however defenseless against
networks [7]; a non convex space of 1070 configurations were
malicious [4]. Thusly, researchers have been concentrating on
efficiently handled to find an optimal configuration.

IJESC, April 2020 1 http:// ijesc.org/

The proposed method is two-fold. trigger the same consequence, and so the proposed method will
1) The proposed strategy effectively chooses unrestorable cutset test only such suspicious cutsets and can ignore the others.
applicants (suspicious cutsets), which incredibly decreases the
quantity of cutsets to be thoroughly tested. The proposed method 2) The proposed method tests the suspicious cutsets rigorously
employs the hitting set enumeration with ZDDs [8] for this and identifies truly unrestorable cutsets. In advance, we built
reduction; this approach finds sets of switches that are closed in two ZDDs, which represent the topological constraints
common in all feasible configurations (Fig. 2). In other words, if (radiality) and the electrical constraints (line capacity and
all switches in the set were to be opened, the network would lose voltage drop). After updating just the topological constraints for
all feasible configurations. This observation leads to our basic each suspicious cutset, the proposed method conducts the
idea; cutting lines near the switches of a hitting set is likely to satisfiability test without re executing complex power flow
calculations required for the electrical constraints.

II. SROCKS

A. Initialization Operation

Individual diversity at the initialization operation is the basis of B. Crossover Operator

genetic diversity in evolutionary processe of SROCKS. If the The optimization ability of genetic algorithm comes from
difference among individuals is little, the advantage are Initial population genetic diversity. Fig.4.The process of crossover
topology and how to select edges for swapping. (a) Initial operator. (a) Step1. (b) Step2. (c) Step3. (d) Step4. (e) Step5. (f)
topology. (b) Select edges for swapping. Three edge-swapping Step6. (g) Step7. (h) Step8. Increase the genetic diversity of the
methods. (a) Method1. (b) Method2. (c) Method3 of crossover population by selecting different individuals as parent samples
operator will disappear soon. And the evolution of species will to generate new off- springs. It has no fixed evolution direction
depend on the mutation operator only. Since the probability of in the SROCKS. Parent topologies generate new children
mutation operator is small, the evolution of species will be close topologies by crossover operator, which obtains a larger solution
to a single state soon, which causes the termination of evolution. space. Thus, the fitness function will search the best solution in a
So as to guarantee that the individuals among population have a larger space. Generally, hybrid administrator retains some
major distinction in the underlying time, we design the portion of father
accompanying instatement activity. For a unique topology of
scale-free WSNs, the area of each node is stationary. We make
the accompanying change under the reason that the degree
conveyance doesn't change. Firstly, we assign a random
probability between [0,1] as Pinit for each individual of each
population. Pin it controls the frequency of edge-swapping on
the original topology. Then for the selected edges, we select one
of the following three swappingmethodsinFig.3.Thus,each
individual transformed from the original topology is completely
random and has a large difference from others. An original
scale-free WSNs topology, and we will do the initialization and mother qualities, and in the long run produces new children
algorithm on it to generate a new topology which is different topologies. The crossover operator in this paper keeps the initial
from the original topology but has the same degree degree of each node unchanged, which means that we will
distribution. Firstly, a random number between [0,1] is preserve the degree distribution of parent topologies Taking into
generated as the edge-swapping frequency Pinit. Then each edge account the communication range in WSNs, we design the
of the topology is traversed one by one. For each edge, Pinit is cross over operator as follows. Suppose ttf and tt mare the
compared with aran dom probability rv between [0,1]. If rvis father’s topology and mother’s topology, respectively. And tts
more than Pinit, we will proceed to the next edge. If rv is less and ttd are the son’s topology and daughter’s topology,
than Pinit, we will do the edge-swapping operation the edges of respectively. Firstly, the parents are chosen by the probability Pc
roof crossover operator in a population. Then tts in her its
topology, the edge-swapping operation for e12 will be aborted,
father’s topology ttf and ttd inherits its mother’s topology ttm.
and we will traverse the remaining edges. Obtain the following sets of edges:

IJESC, April 2020 2 http:// ijesc.org/

 Em=EG−(EG∩EG)(4)
EG={eij |eij∈ttf}(1)
where, EGis the set of the father’s edges, and EG equal to 3
EG={eij |eij∈ttm } (2) before crossover operator, and after the set of mother’s edges.
Efand Em are the sets of father’s exclusive edges and mother’s
exclusive edges respectively. Here ‘exclusive’ means an edge
only exists in one parent’s set but not the other. That is, Ef is
totally different from Em. Because the location of each node
is stationary, if one edge exists in Ef, we also can build it in ttd.
Finally, the son’s topology disconnects the existing edges to
build every mother’s exclusive edges Em, during which the
initial degree of each node is kept unchanged. And the
construction they still keep unchanged. Therefore, it is
process of daughter’s topology is similar to above operation. consistent with the criterion that keeps the initial degree of each
node unchanged. The process that the mother generate sits
Fig. 4 illustrates the process of crossover operator. The
daughters topology is similar to the above operations as shown
representation the connections between nodes in the father’s
in Fig.4 (e-g). Finally, we can see the father’s exclusive edge e12
topology ttfand mother’s topology ttm. It can be seen that the
in the daughter’s topology in Fig. 4(h). Besides, when the
father has an exclusive edgee12 between node 1with node 2 in
father’s topology generates his son’s topology, if the node 3
Fig.4 (f), and mother has an exclusive
cannot find an eligible node to match the node 7 which is the
candidate neighbor of the node 4,
Ef=EG−(EG∩EG)(3)

Figure.5. Candidates for the topology connection. (a) Selected ges eij and ekl. (b) Candidate1. (c) Candidate2

The node 4 will sequentially choose another candidate node in like structure, we generate a new edge between two nodes that
the candidate list. And the node 3 will search all of its neighbors have a similar degree, during which the initial degree of each
for each candidate node until it finds an eligible node to match node is unchanged. We select two edges in the individual
the candidate node. If the node 3 still cannot find an eligible topology and judge the four end nodes of these two edges
node after traversing the candidate list of the node 4, we give up whether they are in the communication range of each other to
generating this edge. guarantee that we can generate ane wedge among these four
nodes. For the edges eij and ekl selected in Fig. 5(a), we propose
C. Mutation Operator a criterion to sort degree and swap edges as follows Eq.(4).
The mutation operator in SROCKS not only increases the Based on the criteria mentioned above, the nodes that have
diversity of a new generation but also produces individuals with similar degrees will connect with each other, thus enabling the
high fitness values. We choose the individual by the mutation evolution of individual topology towards the onion-like
probability Pmut. The goal of mutation operator is to increase structure. Besides, the swapping threshold P swap is defined in
the robustness of the selected individual through exchanging [0, 1), and it cannot be 1 because the two edges will not be
edges, during which the in itialdegree of each node is swapped in that case. We control the efficiency of mutation
unchanged. Metric R is used to measure the robustness of operator by adjusting the value of Ps wap. The appropriate
topology. We search for optimal solution within the local area swapping threshold Ps wap can effectively avoid inefficient
by the mutation operator. Herrmann et al. [8] have found that swapping edges operation.
onion-like structure is stable and robust against malicious
attacks. The nodes with similar degrees connect to each other in D. Migration Operator
onion-like structure. If a node with a large degree failed, another The migration operator in SROCKS is designed to overcome
node with a large degree will replace its function. Therefore, we premature convergence. Individuals with high fitness values can
can minimize the adverse effects of failure nodes as much as be introduced into other populations through migration operator,
possible, and the network topology will remain robust. In order which can effectively prevent falling into a local optimum. We
to make the evolution of individual topology toward the onion- assign different mutation operator and crossover operator

IJESC, April 2020 3 http:// ijesc.org/

probabilities for different populations, and the suitable We simulate SROCKS in MATLAB. The nodes are deployed
individuals are selected in different populations in every randomly in a disk area with a diameter equal to 500m.
generation. Thus, genetic communication can be carried out Considering that each node must have sufficient neighbors in the
between different populations to prevent trapping into local initial topology, the communication range is set to 200m, and
optima. The migration operation is divided into three steps. each node has 2 edges. We set the number of populations Mpop as
Firstly, we select the individual with the highest fitness value in 10, and the number of individuals in a population Nind as 20.
each population. This optimal individual will be temporarily
These two parameters are determined to be optimal for
stored in an elite population. Secondly, the worst individual in
SROCKS by a large number of experiments. degrees connect
each population is selected. Finally, each optimal individual
with each other in Fig. 5(b). Finally, our proposed algorithm
stored in the elite population will be used to replace a worst
makes the network topology close to the onion-like structure.
individual in a different population. Fig. 6 depicts a complete
process of migration operation. Each population has ten
individuals. And each column represents an individual, and its
length indicates the value of the individual fitness function.

III. SIMULATION RESULTS

<Pswap
the best individual of each population and mark it by bright
color in Fig. 5(a). The selected individuals make up the elite
Where in, di, dj, dk, dlare the degree of node i, node j, node k
and node l respectively. We sort them in descending order, and
name them as d1, d2, d3, d4. Pswap controls reduction ratio of
degree difference. If the formula on the left is less than Pswap, we
will swap edges according to d1, d2, d3, d4. There are two
candidate strategies in Fig.5(b)andFig.5(c). population in Fig.
5(b). Then as shown in Fig. 5(b), the worst individual in each
population is identified and marked by red. Finally, in Fig. 5(c),
the current optimal individual of popula- tion 1 is used to replace
the worst individual of population 2. The current optimal
individual of population 2 is used to replace the worst individual
of population 3.

IV. FAST SCREENING ALGORITHM

The proposed method tests the suspicious cutsets rigorously and 200
identifies truly unrestorablecutsets. In advance, we built two
Throughput

ZDDs, which represent the topological constraints(radiality) and 150

the electrical constraints (line capacity and voltage drop). After
updating just the topological constraints for each suspicious 100
cutset, the proposed method conducts the satisfiability test
SROCKS
without re executing complex power flow calculations required
50 ROCKS
for the electrical constraints A formal definition of an
unrestorablecutset is given as follows; given an un restorable
cutset,
0
123456
where G_ is a sub graph without vertices U and their edges (i.e., Number of Populations
we cut sections of U and isolate them by opening neighboring
switches), and the last term means that G_ has no feasible
configuration and cannot deliver power to all intact sections. V. CONCLUSION

Due to the characteristics of WSNs, such as limited

communication range and limited node degree, the traditional
BA model in wired network is no longer suitable. Then, we
proposed a scheme called SROCKS to optimize the robustness
of the constructed scale-free topologies. In this SROCKS, we
designed two novel operators, namely crossover operator and
mutation operator. Under the evolution of these two operators,

IJESC, April 2020 4 http:// ijesc.org/

the initial degree of each node is unchanged, thus the scale-free
property is preserved. Finally, we simulate our algorithm and
two existing algorithms on their performances in improving the
robustness of scale-free WSNs topologies under different edge
densities and network sizes. The experiment results show that
our algorithm can significantly improve the robustness of scale
free WSNs against malicious attacks. SROCKS is designed for
enhancing the robustness of WSNs in a centralized system,
which needs the information of the entire scale-free network
topology. We will further focus on WSNs in a distributed
system, and explore the application of multi-population co-
evolution algorithm in it.

VI. REFERENCES

[1]. J. Yick, B. Mukherjee, andD.Ghosal, “Wireless sensor

network survey,” Computer Networks, vol. 52,no. 12,pp. 2292–
2330, 2008.

[2].L. M. L. Oliveira, J. J. P. C. Rodrigues, A. G. F. Elias, and B.

B.Zarpel˜ao, “Ubiquitous monitoring solution for Wireless
Sensor Networks with push notifications and end-to-end
connectivity,” Mobile Information Systems, vol. 10, no. 1, pp.
19–35, 2014.

[3].C. Komar, M. Y. Donmez, and C. Ersoy, “Detection quality

ofborder surveillance wireless sensor networks in the existence
of trespassers’ favorite paths,” Computer Communications, vol.
35,no. 10, pp. 1185–1199, 2012.

[4]. J.M. Corchado, J.Bajo, D. I. Tapia, andA.Abraham, “Using

hetero generous wireless sensor networks in a telemonitoring
systemfor healthcare,” IEEE Transactions on Information
Technology in Biomedicine, vol. 14, no. 2, pp. 234–240, 2010.

[5]. B. C. Villaverde, S. Rea, and D. Pesch, “InRout-a QoS

aware route selection algorithm for industrial wireless sensor
Networks Tingrui Pei, DaXie, Zhetao Li, Dengbiao Tu,
Youngjune Choi, "A Heterogeneous-aware Cooperative MIMO
Transmission Scheme in WSN", Journal of Software, vol. 8,
no.3, pp. 639-644, 2013.

[6]. Li Yang, DaiYunWeng, "Research of Wireless Sensor

Routing Algorithm Based on Uneven Clustering", Journal of
Networks, vol. 7, no. 3 , pp.584-590, 2012.

[7]. Chunyao FU, Wei WEI, Ang WEI, "Application of

Quadratic-Standard Form Based on LEACH Protocol Analysis",
International Journal of Digital Content Technology and its
Applications, vol. 6, no. 6, pp.16-26, 2012.

[8]. Liao Ming-hua, Zhang Hua and Sun Guang, "EnergyAware

Routing Algorithm for Wireless Sensor Network Based on Ant
Colony Principle", Journal of Convergence Information
Technology, vol. 7, no. 4,pp. 215- 221,2012.