Content uploaded by Nadeem Javaid
Author content
All content in this area was uploaded by Nadeem Javaid on Oct 20, 2021
Content may be subject to copyright.
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
MS Thesis Presentation
Enhancing robustness of scalefree IoT networks again
st random and malicious attacks
Muhammad Usman (MS Scholar)
CIIT/FA18/REE/016/ISB
Supervisor: Dr. Nadeem Javaid (Associate Professor)
Department of Computer Science
CoSupervisor: Dr. Sardar Muhammad Gulfam (Assistant Professor)
Department of Electrical and Computer Engineering
COMSATS University Islamabad, Islamabad 44000, Pakistan
1
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Introduction
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Introduction (1/3)
[1] !"# $"%&% "# '"() *" + * * $" ,)*! .
/
0123 4((562(56.76897 ***$'% ,%! *! IEEE/ACMTransaconson
Networking(TON):::
•Availability of efficient and reliable sensors make them common in our life.
•Wireless Sensor Networks (WSNs) [1] formed by using these sensors.
•A large number of nodes are deployed in WSNs such as [2],
•sensor nodes and
•sink nodes.
•Sensor nodes form a multihop network to share the sensed data with sink node.
•Sensed data can be from,
•healthcare, power grids, agriculture, security automation, industrial
machines, smart homes, etc.
;
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Introduction (2/3)
•Critical and sensitive applications are running over these networks.
•The attacks greatly reduce the robustness of these WSNs [2].
•In terms of target selection there are two types of attacks [3],
•random attacks and
•malicious attacks.
•Some network topologies are resistant to random attacks and some to
malicious attacks.
• The scalefree topology is robust to random attack but fragile to malicious
attack.
• Scalefree topology mainly belongs to the complex network theory [4].
[2] 23 4((562(56.76897 ***$'% ,%! *! IEEE/ACMTransaconson
Networking(TON):::
[3] <='7>3;# *", & "*, ,& # &,! *! 888,& > #&$;;??
?;.
[4] 7 $@@</&,$ *% ,$* * %,%*! " "$,A,& PhysicaA9*&&,,,"*
##,& B;;

Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Introduction (3/3)
• The Scalefree model is the most appropriate and suitable model because
•Fewer high degree nodes
•Large number of low degree nodes
•Homogeneous nodes have,
•similar communications range,
•similar bandwidths and
•similar processing capabilities.
• The onionlike structure is proved to be better solution for network robustness
[5],
•in onionlike structure the highest degree node at center,
•each ring represents the same degree nodes and
•as we move away from center the degree of nodes decrease. Fig. 1 OnionLike Structure
01<,"> ""<<"+6 *! * # $', *$*, C,JournalofStascal
Mechanics:TheoryandExperiment,2011#D.
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Literature review
?
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Technique (s) Achievement (s) Limitation (s)
Simulated Annealing (SA) [6] Robustness optimization of scale
free network Trap into the local optima
Multi Population Genetic
Algorithm (MPGA) [7]
Enhancing the network robustness Poor computation efficiency
ROSE Optimization Algorithm [2] Enhances network robustness
through degree difference and angle
sum operation
•Wrong selection of reference
node
•Redundant operations due to
considering all nodes
[2] 23 4((562(56.76897 ***$'% ,%! *! IEEE/ACMTransaconsonNetworking
(TON):::
[6] =D5 #6#&E$* * %,%*! !**"$*& > %, "#&+")*
> #&$$ *##?..?#$="$
[7] 2@<()$&FEE."*"+ *$ *% *** %*$*,&IEEECommunicaonsMagazine/
;
Literature review (1/5)
.
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Technique (s) Achievement (s) Limitation (s)
ROCKS [8] Enhance the network robustness Not for the distributed system
3step strategy to generate onion
like structure [9]
Give relation between robustness
and community structure Not define different community
subgraphs effect on robustness
Edge classification
[10]
Through edge classification the
robustness of the network improve
In small sized networks its
performance is compromised
[8]2@<((56:7 *6#&E& ,(*&D #& > 8+ & % ,( )*! 888G>,&
)*! $
[9] H$H@3>H3$4($# +$* * %, #I*! !*#+$, *'*,*PloSone?
[10] 7 $@@</&,$ *% ,$* * %,%*! " "$,A,& Physica9*&&,,,"*##,&
;;
Literature review (2/5)
/
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Technique (s) Achievement (s) Limitation (s)
Clustering Routing Scheme [11] Virtual Cluster Head (CH) improve
the robustness
Due to limited energy of sensors
this strategy not work for long
Robustness Link Protection method
RobLPGA based on Genetic
Algorithm (GA) [12]
Robustness of the network is
improved
The addition of link cause extra
cost
Nonlinear Preferential Rewiring
(NPR) [13]
Convert nonscalefree network to
scalefree network
High computational time
[11]@<(>D7> <4:8J,** *7 &$ ( )*! =" =#&* !K# "$IEEEAccess.
;
01DEE&> ,+ /5,K&,$ *% 8,$*7 * %> #I)*! $@D *,& InInternaonalConferenceonComplex
NetworksandtheirApplicaons##/./:#$>
0;1HD@H> /)*! * # $' #&E& '*$ ,%*! * ,%*! $ #%&!$* "
InternaonalJournalofDistributedSensorNetworks.././..
Literature review (3/5)
:
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Technique (s) Achievement (s) Limitation (s)
Self competition among the
individuals [14]
Solve the premature convergence of
GA
Uses multiple operations which
make it difficult for realworld
applications
kcore decomposition [15] Shell based attacks are introduced
against which the robustness is
calculated
Not following onionlike structure,
therefore, not for the SFNs
AI based optimization technique
[16]
Requires less computation cost than
heuristic algorithms
Not suitable for different size of
networks and edge densities
[14]2@3@L4K"(56<'"#&+7 *8+ & $ *!*%> #&& % ,** %$888)6>6888
> %, > #*> ,& ##?
01@4($@"4>)!@C,**$ %> #I)*! =" > 5, # & 888,& >,*"'*98I#
=%?.##;.;?
0?1>)23 4@L"&FEE:*$* **! $,% *** %$888> ,& $E.
##::
Literature review (4/5)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Technique (s) Achievement (s) Limitation (s)
Multi Agent GA (MAGA) [17] Optimal topology is found by the
MAGA using multiobjective
optimization
A single measure needs to be used
for these objectives
A better network structure required
that is robust against attacks
Degree and Betweenness (DB)
based model [18]
Node importance is calculated
based on its DB
Removal of important node causes
other nodes and edges to become
overload
Load is distributed among
neighboring nodes high load
Simultaneous edge and node
removal is not discussed
PROSE [19] Protect the network from the
attacks by installing backups
The backup nodes have the same
characteristic, therefore, can be
easily removed
0.1@3"@<?&$*$&,$ *% , *'"*,& , #I*! PhysicaA:StascalMechanicsanditsApplicaons449##;;?;.
0/1<"<3:>,"$% " %,%*! % + "$"$%PeertoPeerNetworkingandApplicaons12?##?.?;.
0:1%M/D768ND ,&+7,** %$9$*"C,"> *IEEESensorsJournal18##:.
Literature review (5/5)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Motivation and problem statement
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Motivation
•WSNs in IoT are part of multiple and critical fields [7, 20] like, healthcare,
transportation, agriculture, power grid systems, etc., the attack on the network
causes a lot. Taking it as a motivation, we are working to find a solution in
which the malicious attacks can efficiently tackled, and the network robustness
being improved.
0.12<@()$&FEEO5*5+7 *$ *% *** %$*>&P888> ,& $E.
0125@ 4)5 *(" O$"' "!*! "% # +$* * %* $ ** %$P> #*)*! ?
;
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Problem statement
Due to the malicious attack on the network, it is divided into multiple subgraphs.
By using the rewiring technique, authors in [6] increase the number of nodes in
maximum connected subgraphs. However, instead of finding the global solution the
model traps into the local optima.
To find the global solution Qie et al. [7] give the MPGA technique because the
conventional GA has premature convergence multipopulation can resolved this.
The Qie et al. [2] enhance the robustness of the scalefree network by angle sum
and degree difference operations. The robustness of the network is enhanced,
however, the performed operations increases the complexity, and the wrong
selection of reference node are two major issues.
[2 ] 23 4((562(56.76897 ***$'% ,%! *! IEEE/ACMTransaconsonNetworking
(TON):::
[6] =D5 #6#&E$* * %,%*! !**"$*& > %, "#&+")*
> #&$$ *##?..?#$="$
[7] 2@<()$&FEE."*"+ *$ *% *** %*$*,&IEEECommunicaonsMagazine
/;

Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Contributions (1/3)
Improved ScaleFree Networks (ISFNs) strategy
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Network model description
Contributions (1/3)
?
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Network model description ( 1/6 )
•ScaleFree Network construction
•Scalefree network is constructed by using the
famous BarabásiAlbert (BA) model improved in [2].
•The model start with a small clique.
•The network growth follows preferential attachment.
•The probability for a new node i is calculated
by using Equ. 1,
• , . . . (1)
•where, and are degree of nodes i and sum of degrees of neighboring nodes,
respectively.
•
Fig. 2 Connection Process of the Nodes
[2] 23 4((562(56.76897 ***$'% ,%! *! IEEE/ACMTransaconsonNetworking(TON):
::
Contributions (1/3)
.
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
•Independent edges
•The Improved ScaleFree Network (ISFN) consists of two operations:
•Edge’ degree based swap and
•Nodes’ distance based swap.
•Before explaining the above operations, we need to know about the independent
edges.
•The scalefree network topology consists of the graph G = (N, E),
•where N is the number of nodes and
•E is the number of edges.
•The edges eij and ekl independent if they satisfy these two conditions,
•nodes i, j, k and l must be in the same communication range. Through this they can make connection
with each other and
•between nodes i, j, k and l there is no edge except the eij and ekl. Independent edges swap is presented
in the Fig. 3.
Network model description ( 2/6 )
Contributions (1/3)
/
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
•Independent edges
•In Fig. 3a the original topology
is shown.
•We have edges eij and ekl .
•In Fig. 3b and 3c the edges are
swapped and we have eik , ejl and
eil , ejk, respectively.
:
Fig. 3 Independent Edge Swap a: Original Topology
b: First Edge Swap c: Second Edge Swap
Network model description ( 3/6 )
Contributions (1/3)
:
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Metric of robustness
•A robustness metric (R) based on the percolation theory proposed by Schneider et
al. [21].
•In this metric, removal of high degree nodes, divides the network into subgraphs.
•The robustness is the number of nodes in the Maximum Connected Subgraphs
(MCS)
•The robustness metric R is defined as,
•where, N is the total number of nodes in a network,
1/ N + 1 is a normalization factor,
MCSn is maximum connected subgraphs nth high degree node removed and
summation means nodes removal after each attack is considered
•The robustness value lies in the range of (0, 0.5).
01,"> ""<
+"<&$& %
, C, *! Proceedingsofthe
NaonalAcademyofSciences,108##;/;/;/
. . . (2)
Network model description ( 4/6 )
Contributions (1/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
•Edges degree based swap
•The edge degree is derived from the nodes degree. The edge degree dij of the link eij is
defined as
. . . ( 3 )
•where di and dj represents the degree of node i and j, respectively.
•Larger the value of the dij the more important the edge.
•To perform the edge degree based swap we first find the independent edges and then
calculate the degree of these edges by using Equ. 3.
•After that we calculate the edge degree difference by using the Equ. 4, 5 and 6.
•Then a pair of edges are selected for which the degree difference is minimum.
•If the new edges not destroy the initial topology and not decrease the robustness the swap
is accepted.
. . . ( 4 )
. . . ( 5 )
. . . ( 6 )
•
Network model description ( 5/6 )
Contributions (1/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
•Nodes' distance based edge swap
•Second part of our ISFNs technique is the nodes’ distance based edge swap.
•A minimum distance from one node to the other node is found.
•In this approach, we first find the distance for edge pairs eij and ekl then find the
average of the obtained eccentricities.
•We calculate distance for all the possible edges swap and take their average.
•The pair of edge with minimum value of distance is selected.
•If the swap edge improves the robustness, we keep the edge swap otherwise
keep the original edge.
Network model description ( 6/6 )
Contributions (1/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (1/8)
•The experiments are performed using MATLAB.
•We used topologies having a different number of nodes and compare the result of the optimized
topology with the existing techniques such as SA and ROSE.
•IoT network is deployed in circular field for better economy [22].
Parameters Values in ROSE, SA and ISFN
Number of nodes (N) 100  300
Average of simulation results (Avgrun) >25
Diameter of the field (Fielddia) 500m
Range (sensor) 200m
Neighbor threshold (neighourthreshold) 25
01D"+8>)= ,O>,>&9##$I>&& P8+ * +& " ,*& Q ?,/
Contributions (1/3)
;
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Comparison of ISFNs with ROSE and SA
Simulation results (2/8)
•All the techniques improve the robustness of the network as
compared to the initial network.
•The high robustness of ISFNs is due to the formation of
better onionlike structure.
• Moreover, the operations as part of the ISFNs reduce the
redundant operations to optimize the network against
malicious attacks.
•From the results, it can be concluded that the networks are
initially robust and the optimizations enhanced the
robustness.
Fig. 4 ISFNs evaluation of robustness
•The minimum value of SA is due to the local optima problem.
•Moreover, due to the possible wrong selection of the centre node and the redundant operations
caused by the degree difference and angle sum operations reduced the performance of ROSE.
Contributions (1/3)

Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (3/8)
Comparison of ISFNs with ROSE and SA against low degree attacks
•All the techniques have downward trends due to the
low degree nodes removal.
•As the SFNs are robust against the low degree nodes
attacks.
•Therefore, the network robustness decreases gradually.
•The high robustness against the low degree attacks
confirm that the network remains scalefree after
optimization.
•Network robustness against low degree attacks is
shown in Fig. 5.
Fig. 5 Network performance evaluation
against low degree attacks
Contributions (1/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (4/8)
Comparison of ISFNs with ROSE and SA against high degree attacks
•The robustness of the network decreases sharply after the high
degree nodes are removed.
•All the techniques have the downward trends, however, due to
the change in robustness value, the schemes could not be
compared directly.
•Therefore, against the number of removed nodes, the slope is
analysed.
•As in Fig. 6, the existing techniques and the initial network
collapsed after the removal of almost 20 nodes.
•However, the proposed ISFNs scheme performs much better and
the network fragments after the removal of 40 nodes.
Fig. 6 Network’s performance evaluation
against high degree attacks
•The high robustness value ensures that the network is optimized with the better formation of
an onionlike structure.
•The two operations introduced in ISFNs proved their usefulness by providing a robust
network structure against malicious attacks.
Contributions (1/3)
?
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (5/8)
Comparison of ISFNs with ROSE and SA against random attacks
•After each attack, the techniques have downward trends because the
nodes are removed from the networks, as shown in Fig. 7.
•The robustness of ISFNs outperform the existing schemes and initial
network, due to better network structure.
•Almost 80% of the nodes are required to be removed from the
network to make it collapse.
•However, the existing schemes and the initial network resist only
against the removal of 60% of the nodes.
•The small rise in the ROSE during the 20% and 35% of the nodes
removal is due to the random selection of high degree edges during
the degree difference operation.
Fig. 7 Network’s performance evaluation against
random attacks
•SA has less robustness as compared to ROSE and ISFNs because in the network evolution process the
constraint of sensor nodes such as communication range and the threshold value of nodes degrees are not
considered, therefore, the removal of high degree nodes fragment the network.
Contributions (1/3)
.
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (6/8)
Comparison of ISFNs with ROSE and SA against variable attacks
•When the number of removed nodes is small, the maximum number of nodes is present in the MCS.
•SA and ROSE performed worst, however, ISFNs outperform the existing techniques.
•During the simulations, 3 different numbers of nodes are randomly generated and against all these,
ISFNs has better results.
•So, these results prove the importance of edges' degrees and node's distance based swaps operations to
form the onionlike structure.
•Moreover, against the multiple
nodes' removal from the
network, robustness is
calculated and the number of
removed nodes is randomly
selected.
•The network performance
against the variable attacks is
shown in Fig. 8. Fig. 8 Network’s performance evaluation against variable attacks
Contributions (1/3)
/
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (7/8)
Comparison of network robustness against different number of nodes
•The increase in the network size causes the robustness of the
network to decrease, as shown in Fig. 9.
•It is due to the availability of a large number of high degree
nodes within the same sensor field.
•Therefore, the removal of high degree nodes has a severe
effect on the network connectivity.
•All the techniques have the same downward trends with the
increase in the number of nodes.
•However, ISFNs outperforms the existing techniques for
different number of nodes.
•It is due to the better onionlike structure formed by the
ISFNs operations.
•From the results, it is shown that the proposed scheme is
better, when it is implemented on the dense and complex
networks.
Fig. 9 Network’s performance evaluation against different
number of nodes
Contributions (1/3)
:
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
N Algorithms Best Average Worse
100
SA
ROSE
ISFNs
0.1200
0.2120
0.2620
0.109
0.177
0.227
0.0980
0.1420
0.1920
150
SA
ROSE
ISFNs
0.0893
0.1300
0.1853
0.078
0.11935
0.172
0.0667
0.1087
0.1587
200
SA
ROSE
ISFNs
0.0570
0.1180
0.1730
0.0525
0.1065
0.159
0.0480
0.0950
0.1450
250
SA
ROSE
ISFNs
0.0832
0.1204
0.1688
0.078
0.1114
0.1608
0.0728
0.1028
0.1528
300
SA
ROSE
ISFNs
0.0487
0.0653
0.1120
0.0437
0.0613
0.1096
0.0387
0.0573
0.1073
Simulation results (8/8)
Comparison of ISFN with SA and ROSE for Different Number of Nodes
Contributions (1/3)
;
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Contributions (2/3)
Network Topology Evolution Scheme (NTES)
;
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Scalefree model
;
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Scalefree model (1/2)
BarabasiAlbert model in wireless sensor network
•Realworld networks such as hospitals, military, transportations, etc., are ScaleFree Networks
(SFNs)
•The operations of dense networks is difficult
•The small network's operations are easy
• Failure, of these networks has less effect on the performance
of the whole network
• So, the whole network is the sum of its parts.
•The dotted line represents the partition of the network and Nu and Ni are the center nodes
•The network growth starts from these center nodes.
• The networks are links by onetomany correspondence.
•To increase the robustness of the network edge swap is performed.
;;
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Scalefree model (2/2)
Types of edge swap
•We have categorized the edge swap into three categories;
•Edge swap of randomly selected nodes
•Edge swap of degree based selected nodes
•Edge swap of distance based selected nodes
•In random edge swap, the nodes are randomly selected
•In degreebased edge swap the nodes are selected based on degree
•The long links are added by nodes’ distancebased edge swap
;
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview
;
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview (1/7)
Network topology evolution
•A network having a smaller number of nodes is more robust
against malicious attacks [2, 8, 23].
•Therefore, to generate a robust network topology, nodes are
distributed equally into two parts, as shown in Fig. 11.
•In each part, the network is evolved by considering the
powerlaw distribution.
•The connection of both networks is made by onetomany
modes of connection.
•In onetomany correspondence, the high degree nodes of
one part connect with low degree nodes of other part of the
network.
;?
[2] Tie Qiu, Aoyang Zhao, Feng Xia, Weisheng Si, and Dapeng Oliver Wu. Rose: Robustness strategy for scale free wireless sensor networks. IEEE/ACM Transactions on Networking, 25(5):2944–2959, 2017.
[8] Tie Qiu, Jie Liu, Weisheng Si, and Dapeng Oliver Wu. Robustness optimization scheme with multipopulation coevolution for scalefree wireless sensor networks. IEEE/ACM Transactions on Networking,
27(3):1028–1042, 2019.
[23] Mingxing Zhou and Jing Liu. A memetic algorithm for enhancing the robustness of scalefree networks against malicious attacks. Physica A: Statistical Mechanics and its Applications, 410:131–143, 2014.
Fig. 11 Network topology evolution with powerlaw distribution
and onetomany correspondence
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview (2/7)
Network topology evolution
•The dotted line in Fig. 11 shows the division of the network.
• In both parts, equal number of nodes are deployed.
•The blue nodes are used to denote network A (NA), black
nodes are used for network B (NB) and NM denotes the
nodes that connect both networks.
•The black solid lines denote connectivity links (CL) and
double lines are for the mutual links (ML).
•Both upper and lower parts form a network by
synchronously adding edges for each node.
;.
Fig. 11 Network topology evolution with powerlaw
distribution and onetomany correspondence
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview (3/7)
Nodes’ degree distribution based on kcore
•The onionlike structure consists of rings of different degree nodes.
•Each ring represents nodes of the same degree [8].
•Malicious attacks on high degree nodes required high computational cost.
•Therefore, finding the information about the nodes' degree to affect a specific part of the
network has less computational cost.
•So, we propose a method to find the degree of nodes in each ring and remove a node based
on its importance.
;/
[8] Tie Qiu, Jie Liu, Weisheng Si, and Dapeng Oliver Wu. Robustness optimization scheme with multipopulation coevolution for scalefree wireless sensor networks. IEEE/ACM Transactions on
Networking, 27(3):1028–1042, 2019.
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview (4/7)
Nodes’ degree distribution based on kcore
•In each ring same degree nodes are presented.
•These rings are created through kcore decomposition.
•In kcore decomposition, nodes' removal start from low
degree nodes and placed in C4.
•The degree is recalculated and low degree nodes placed in
C3.
•Process continues until the highestdegree nodes are
removed.
•In Fig. 11, C1 is the highest core.
;:
Fig. 11 Network topology evolution with powerlaw distribution
and onetomany correspondence
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview (5/7)
Attacks on the designed topology
•The attackers have complete network information to paralyze
it.
•So, the defender should take measures against the attacks.
•Therefore, two types of attacks: inner core nodes and nodes'
degreebased attacks are considered.
•In a corebased attack, inner core nodes are removed because
they have more influence.
•A network with the core of high degree nodes is shown in
Fig. 12 and red color nodes (NR) are removed.
•Initially, by removing high degree nodes the network is still
connected.

Fig. 12 Attacks based on inner core nodes
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview (6/7)
Attacks on the designed topology
•After performing malicious attacks on the core nodes, the
network fragments into multiple subgraphs Fig. 13.
•The high degree nodes present in the core of the network are
removed. These nodes are shown in red color NR in Fig. 13.
• S1, S2 and S3 are the subgraphs that are made after core
based attacks.
•To more effectively destroy the network, high degree nodes
in the subgraphs are removed.
•These nodes are presented as NMCS.

Fig. 13 Attacks based on high degree nodes that are part of
MCS
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
NTES overview (7/7)
Robustness optimization of the NTES using heuristic algorithms
•To optimize the SFNs three heuristic algorithms GA, ABC and BFO are used.
•In GA, the edge swap is made by considering the exclusive edges [7].
•ABC and BFO algorithms require random position change to find the global optimal solution.
•In SFNs, a random position change is not possible; therefore, a degreebased edge swap and a
random edge swap are made.
•When the operators in these algorithms are improving robustness then a degreebased edge
swap is performed; however, when they trap into local minima random edge is performed.
•In other words, the exploration is performed by random edge swap and degreebased edge
swap is used for exploitation.

[7] Tie Qiu, Jie Liu, Weisheng Si, Min Han, Huansheng Ning, and Mohammed Atiquzzaman. A datadriven robustness algorithm for the internet of things in smart cities. IEEE Communications
Magazine, 55(12):18–23, 2017.
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Details of NTES Algorithms
;
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Details of NTES Algorithms
(1/3)

Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021

Contributions (2/3)
Details of NTES Algorithms
(2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
?
Contributions (2/3)
Details of NTES Algorithms
(3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results and discussion
.
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Network topology evolution
•The nodes are randomly deployed in the sensor field of 500*500m2
•The total number of nodes is 100
•The communication range of the nodes’ is 50% of the total size of the sensor field
•Since our work is to make small networks; therefore, the nodes are equally distributed in both
parts
•Due to the nodes’ resource constraint, the minimum and maximum value of nodes’ degree is set to
be 2 and 25, respectively
Simulation results and discussion (1/9)
/
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Attacks on upper and lower networks
•In random attacks, nodes are removed randomly
•Malicious attacks happen on high degree node
•Two nodes are removed simultaneously in each attack to analyze the
network
•Initially, the value of robustness remains the same for both attacks
•The malicious attack is more severe than the random attack. When 15
pairs of nodes are removed from the network
•In Fig. 14, attacks effects on upper and lower network are shown.
•The difference in robustness is small between the random and
malicious attack that proves the effectiveness of the proposed
technique
Simulation results and discussion (2/9)
:
Fig. 14 Random and malicious attacks on upper and lower
network
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Networks connection by onetomany correspondence
•The powerlaw distribution of the mutual nodes of the network is
validated.
•The powerlaw for the mutual nodes is shown in Fig. 15.
•As per the definition of the powerlaw, there are few number of nodes
having a high degree as compared to the number of low degree nodes.
•The results prove that the low degree nodes are more as compared to
the highdegree nodes.
•Moreover, due to the predefined limit of the nodes that can connect
with the other part of the network, only few nodes are in the mutual
part of the network.
Simulation results and discussion (3/9)
Fig. 15 Powerlaw distribution of the mutual nodes
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Random and Malicious attacks on the network
•Random and malicious attacks are performed on the
topologies that are optimized by GA, ABC and BFO in
Fig. 16.
•ABC performs the poor exploitation on the onlooker bee
phase.
•BFO and GA performs the best against the random attacks.
•However, the computational cost of BFO is high as
compared to GA.
•On average, the optimal solution provided by GA is better
than the one provided by BFO.
Simulation results and discussion (4/9)
Fig. 16 Malicious attacks on the network
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Corebased attack on the network
•The corebased attack is performed in Fig. 17
•The removal of nodes starts from the center core
•Then the lowdegree cores are removed and the
robustness is calculated.
•Initially, same decreasing trend of network robustness is
observed for both degreebased and corebased attacks.
•This is due to the fact that high degree nodes are
presented in the core
•The degree is recalculated on every removal of nodes;
however, the cores are only defined once
•Due to the presence of a large number of lowdegree
nodes, the robustness becomes stable from 40 to 60
nodes against corebased attack.
Simulation results and discussion (5/9)
Fig. 17 Comparison of core based attacks and high degree
node attacks
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Comparison of the robustness for different optimizing
techniques
•The comparison of different algorithms in optimizing NTES is
presented in Fig. 18.
•All the techniques have the same number of nodes and edge
densities.
•The initial NTES topology has low robustness among all the
optimized algorithms that confirms that the optimization greatly
improves the robustness of the network.
•GA outperforms ABC and BFO by enhancing the robustness by
11.33% as compared to the initial network.
•Afterwards, BFO and ABC increase the robustness by 10.6% and
9.24%, respectively.
•GA performance is better; therefore, in the optimization of the SFNs
it is preferred.
•During the optimization the initial network degree distribution is not
changed; therefore, no extra cost is required.
Simulation results and discussion (6/9)
;
Fig. 18 Comparison of optimization algorithms when
N = 100
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Comparison of NTES with different algorithms on the SFNs
•In Fig. 19, the NTES is compared with BA and HC.
•GA R is the robustness of the optimized NTES by GA.
•The highest value of robustness is achieved by optimizing the
NTES with GA. The GA R is higher because the perfect onion
like structure is achieved.
•GA R has 11.36% high robustness as compared to the NTES;
whereas it outperformed BA and HC by 46.90% and 57.08%,
respectively.
•The highest robustness is due to the perfect network topology
construction in which same degree nodes are connected with each
other to form the perfect onionlike structure.
•Moreover, the construction of a network by joining the small
sized networks clearly has more robustness as compared to the
largesized network.
•The smaller networks are more robust, easy to control and easy to
implement.
Simulation results and discussion (7/9)

Fig. 19 Comparison between NTES and existing
algorithms when N = 100
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Comparison of NTES with different algorithms on the SFNs
•The random and malicious attacks are performed on the
networks, optimized by different techniques including BA, HC
and NTES.
•The effect of the random attack is shown in Fig. 20.
•The NTES optimized by GA outperforms all other techniques
while BA performs the worst.
•The techniques follow the downward trend because the removal
of nodes from the network decreases the robustness of the
network.
Simulation results and discussion (8/9)
Fig. 20 Comparison of NTES with existing algorithms
when random attacks happens
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Comparison of NTES With different algorithms on the SFNs
•The comparison of malicious attacks on the robustness of
different techniques is presented in Fig. 21.
•The SFNs are more vulnerable to malicious attacks; therefore,
the network collapses more quickly as compared to the random
attacks.
•The NTES optimized by the GA has outperformed all existing
techniques that is due to the fact by removal of few high degree
nodes the network still connected.
•More attacks are required to divide the network into subgraphs.
•The networks optimized by BA and HC collapse soon due to
not considering the threshold of nodes degree, the
communication range of nodes and trapping into local optima.
•The difference of robustness between BA and HC is small as
compared to NTES and the optimized NTES.
Simulation results and discussion (9/9)
?
Fig. 21 Comparison of NTES with existing algorithms
when malicious attack happens
Contributions (2/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Contributions (3/3)
Optimization of scalefree networks
.
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Overview of JAYA
/
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Overview of JAYA
•JAYA algorithm proposed by R. Rao JAYA is a Sanskrit word that means victory.
•The name is given to the algorithm because in each iteration, it improves the best individual
along with the worst one more efficiently than the other optimization algorithms.
•It is a parameterless algorithm, therefore, no algorithmspecific control parameters are required
that make it easy to implement.
•Moreover, it is a population based algorithm and the individuals are selected based on the
fitness values and are updated as in Equ. 7.
•where, A(i,j,k) is the candidate solution i, j and k represent that in an ith iteration the value of jth
variable of the kth individual, respectively. The A(i,j,best) and A(i,j,worst) are the best and worst
individuals in the search space, respectively.
•The random numbers r1 and r2 are used to generate the diversity in the population and avoid the
solution to stuck into the local optima. The value of these numbers lie between 0 and 1.
•In JAYA, at each iteration, each individual in the population is updated.
•After that, the individuals are compared with their previous values and according to the
required optimization, they are updated.
:
R.
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
JAYA for the SFNs
?
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
JAYA for the SFNs (1/4)
•The SFNs is optimized by JAYA algorithm.
•It is a population based algorithm, therefore, for each topology,
an adjacency matrix (A) is constructed and converted to a
binary coded chromosome as shown in Fig. 22.
• After each iterations, individual is updated based on the best
and the worst characteristics of the individuals present in the
population.
•It is achieved by making the exclusive edges of one individual
into the other individuals.
•In Fig. 22, the topology of four nodes i, j, k and l forming a
complete network is shown.
•The adjacency matrix of the topology consists of binary
numbers i.e., aij = 1, if node i is connected with node j and vice
versa.
•In the same way, if aij = 0 then no link is present between the
nodes i and j.
?
Fig. 22 Chromosome is obtained from the
adjacency matrix
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
•Only the upper triangle is considered to form the chromosome because it has the complete
network information and requires less memory.
•The Equ. 8 is obtained by modifying the Equ. 7, to get the best results.
•where, Ai is the adjacency matrix of the ith topology, whereas Abest and Aworst are the best and worst
topologies, respectively, present in the population based on fitness value.
•r is the number of exclusive edges required to update the ith topology and r is selected randomly.
When the operations are performed on the current topology, the updated form is presented as
Ai+1.
•The adjacency matrix of current topology Ai is
presented in Fig. 23(a) while the best topology Abest in the
population is given in Fig. 23(b).
•The difference between the topologies is calculated by
considering the exclusive edges. As the exclusive edges
are defined as the edges which are presented only in
one topology.
JAYA for the SFNs (2/4)
?
Fig. 23 JAYA for the optimization of SFNs (a) Current individual
(b) The best individual (c) Finding the neighbors
(d) Selecting the nearest node (e) The updated individual
R/
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
JAYA for the SFNs (3/4)
•To enhance the robustness, the exclusive edges present in the Abest needs to be made in Ai.
•In Fig. 23(c), the neighbor nodes are found to make an edge of the Abest into the Ai. Here, the
dotted circle represents the neighbors of node 4.
•The green dotted lines show the distance of node 4 with two of its neighboring nodes.
•The nodes that are in the same communication range and have independent edges as shown in
Fig. 23(d) are selected to make the exclusive edge in the Ai.
•Nodes 3, 4, 5 and 6 are in the same communication range and have independent edges, so in the
edge swap, these nodes are considered.
•The exclusive edge that is part of the Abest is made in
the Ai and termed as Xexclusive1.
•The same method is repeated for the worst and current
individuals and represented as Xexclusive2 After making
these exclusive edges the Equ. 8 become,
?;
R:
Contributions (3/3)
Fig. 23 JAYA for the optimization of SFNs (a) Current individual
(b) The best individual (c) Finding the neighbors
(d) Selecting the nearest node (e) The updated individual
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
JAYA for the SFNs (4/4)
•The process of exclusive edge is repeated for the Xexclusive1 and Xexclusive2 and the Equ. 10 is updated
to the following form.
•Here, XexclusiveT is the adjacency metric obtained by exclusive edges of the best and worst
topologies with the current topology.
•At that point, the two matrixes are adding so, the exclusive edges are made to obtain the Ai+1.
•The process is repeated for all the individuals of the population.
•After the completion of one iteration, the fitness value of Ai+1 is compared with the Ai.
•If the Ai+1 has more fitness than the Ai, the population is updated with the individual having the
highest robustness.
?
R
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Algorithms
?
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
??
Contributions (3/3)
Algorithms (1/2)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
?.
Contributions (3/3)
Algorithms (2/2)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Effects of edge swap on topological
parameters
?/
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Effects of edge swap on topological parameters (1/2)
The following topological parameters are studied to prove the efficiency of these edge swaps.
•Global communication efficiency
•Average clustering coefficient
•Average shortest path length
•Assortative coefficient
Global communication efficiency
A network measure that describes the efficient exchange of information in the network. It is defined as
follows.
where, 1/N(N  1) is the normalization factor and dij is the shortest path between node i and j. Higher the value
of global communication efficiency makes the network more efficient.
Average clustering coefficient
In a network, the clustering coefficient Ck defines the node's characteristics to form a cluster. It is defined as
follows.
where, Ek and Nk are the edges between the nodes and the total number of nodes, respectively, that have degree
k . The value of Ck lies in the range of [0, 1].
?:
R
R
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Effects of edge swap on topological parameters (2/2)
The overall clustering coefficient ¯C¯ of a network is calculated as
where, kmax is the maximum value of the node degree.
Average shortest path length
It is the length between all pairs of nodes. To calculate the average shortest path length D(G), the following
equation is used.
Assortative coefficient
•Assortativity is defined as the links between nodes based on properties like degree, betweenness centrality,
etc. It has a value in the range of [1, 1].
•1 means that the network is highly assortative, 1 proves that the network is disassortative and 0 means
nodes have no.
•The assortative coefficient is calculated as,
•Where, N is the total number of nodes, ki and kj are the degree of node i and j, respectively. M is the total
number of edges.
.
R;
R
R
R?
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Node's addition in maximum connected
subgraph
.
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Addition of nodes in maximum connected subgraphs
•The malicious attacks fragment the network into multiple subgraphs.
•The number of nodes in each subgraph is not the same, some have a high number of nodes and
some have less.
•To get a high number of nodes in the MCS, the nodes present in the smallsized subgraphs
broadcast a request message to the nodes present in its neighboring largesized subgraphs.
•If the degree threshold of the nodes is not achieved the nodes present in the larger subgraphs
receive the request message and reply with a confirmation.
•After the links are formed between the nodes of smallsized and largesized subgraphs, the
robustness is calculated.
•Due to an increase in the number of nodes in the MCS, the robustness of the network is
increased.
•The communication between the nodes is performed through a wireless medium, therefore, no
extra cost is added in the form of adding the edges.
.
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Effects of bridge edge on the network
.;
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Effects of bridge edge on the network (1/2)
•In the networks, some edges are important for the network connectivity because the removal of
these edges effects the network structure.
•These edges are termed as the bridge edges and they are protected to maintain the network
connectivity.
•These specific edges are initially found in the network by continuously removing the edges from
the network.
•An edge is marked as a bridge edge if its removal fragments the network into multiple subgraphs.
•The nodes in the neighbors of the bridge edge are selected and the edge swap is performed
against the nodes that form the bridge edge.
•The procedure to enhance the network robustness with the protection of bridge edge is shown in
Fig. 24.
.
Fig. 24 Effects of bridge edge on the robustness of SFNs
(a) Bridge remove (b) Bridge edge protected with swap
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Effects of bridge edge on the network (2/2)
•In Fig. 24, edge eki is the bridge edge and is removed from the network.
•After the removal, the neighbors of node k and i are selected.
•The edges ekl and eij are independent edges, therefore, the edge swap is performed.
• The edge swap increases the network robustness by increasing the number of edges in the MCS.
.
Fig. 24 Effects of bridge edge on the robustness of SFNs
(a) Bridge remove (b) Bridge edge protected with swap
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Effects of different attacks strength on the network
In the attacks’ strength, specific nodes are removed in an instance. To evaluate the network
robustness, the attacks of the following strengths are implemented.
•Attack strength 1
•Attack strength 2
•Attack strength 3
In the attack strength 1, 5% nodes are removed from the network.
These nodes are selected randomly and after each removal of nodes, the robustness is calculated.
After multiple attacks, the network is analyzed.
Moreover, 10% and 15% nodes are removed in the attack strength of 2 and 3, respectively.
The higher the attack strength paralyzes the network more efficiently.
.?
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results
..
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (1/16)
•In the simulations, the number of nodes is varied from 100350 and are uniformly random
deployed in the sensor field of 500*500m2.
•The communication range of nodes is 50% of the total sensor field.
•The threshold of nodes' degrees is set to 25, when the number of nodes is 100.
•However, when the nodes are increased by 50 the threshold of degree also increases by 5.
•For the JAYA algorithm, the population consists of 30 individuals and 100 iterations are
performed.
•Next, the topological parameters against the edge swaps are analyzed.
•Afterwards, the robustness for the addition of nodes in MCS, the network connectivity against
the bridge edge and the effects of attacks strength are calculated.
./
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (2/16)
•Comparison of JAYA with the existing
•The performance of JAYA is compared with two existing techniques, ROSE and SA for the different
number of nodes, as shown in Fig. 25.
•The maximum value of robustness RJaya is
obtained from the network optimized by JAYA.
•It beats the robustness of ROSE RROSE and SA RSA for the
different number of nodes.
•However, with the increase in the number of nodes, the robustness
decreases.
•Initially, the robustness for the smallsized networks is high because of
the better network structure.
•However, as the number of nodes is increased in the network, it
causes the maximum degree of nodes to increase. As the power law is followed in the SFNs, therefore,
the number of high degree nodes remains less.
•The removal of these nodes fragments the network more severely than the network with less number of
nodes.
.:
Fig. 25 Effects of edge swap on the robustness of SFNs
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (3/16)
/
•Comparison of JAYA with the existing
•The random attack on the initial and optimized topology is shown in Fig. 26.
•The optimized topology has more robustness as compared to the initial topology.
•The nodes removal decreases the robustness, however, it has less
effects on the optimized topology.
•The initial network is collapsed, when 50% of the nodes are
removed from the network.
•However, the optimized topology is survived until 80% nodes
are removed from the network.
•It proves the effectiveness of the optimization technique.
Fig. 26 Effects of random attacks on the robustness of SFNs
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (4/16)
/
•Comparison of JAYA with the existing
•As the malicious attacks happen on the important nodes, based on the degree and SFNs have less number of
high degree nodes.
•Therefore, the network is fragmented, after the removal of a small
number of high degree nodes.
•The optimized topology reduces the effects of malicious attacks
due to better network structure.
•The malicious attacks effects on robustness is shown in Fig. 27.
Fig. 27 Effects of malicious attacks on the robustness of SFNs
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (5/16)
•Analysis of edge swap methods
•The network performance is analyzed against attacks and topology parameters.
•The malicious attacks happen on the high degree nodes and for all
the edge swaps methods the same network topology is considered,
therefore, the degree distribution is not changed.
•Moreover, as the network's degree distribution remains the same
after all types of edge swaps so, the effects of malicious attacks
also remain the same.
•The network obtained after random, degree and distance dependent
edge swaps are represented as Nr, Ndeg and Ndis, respectively, as
shown in Fig. 28.
•Initially, the robustness is high in all the cases, however, as the number of removed nodes is increased the
curves follow the downward trend.
•As the SFNs have a small number of high degree nodes, therefore, the network collapse after the malicious
attacks.
/
Fig. 28 Malicious attacks after the edge swap
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (6/16)
•Analysis of edge swap methods
•In the low degree attacks, the low degree nodes are removed from the network.
•As the SFNs are robust against the low degree attacks, therefore, the
removal of nodes decreases the robustness gradually.
•The low degree attack is shown in Fig. 29.
•All the three edge swaps follow the same downward trend because they
are performed on a network with same degree distributions.
•Moreover, after the attacks, the gradual decrease in robustness shows the
presence of a large number of MCS in the network.
/;
Fig. 29 Low degree attacks after the edge swap
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (7/16)
•Analysis of edge swap methods
•As in random attacks, the nodes are randomly removed from the network.
•Therefore, the effects on robustness for these edge swaps are different,
as shown in Fig. 30.
•As compared to the malicious and low degree attacks, the networks
robustness against the random attacks is different.
•The techniques have the downwards trends, however, the number of
connected subgraphs are more then the malicious attacks.
/
Fig. 30 Random attacks after the edge swap
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (8/16)
•Analysis of edge swap methods
•In Fig. 31, the assortativity coefficient of the networks obtain random ACr, degree ACdeg and distance Adis
dependent edge swap is shown.
•The generated SFNs have disassortative nature.
•The random edge swaps do not affect the assortativity of the network
because in these edge swaps the information of nodes degree is not
considered.
•Therefore, the network remains disassortative.
•As in the distance dependent edge swap, the nodes that are forming long
links are connected.
•So, the network should become more disassortative, however, the results
in Fig. 31 prove that it does not affect the assortativity of the network.
•Moreover, the degree dependent edge swap considers the degree of nodes and connects the similar degree
nodes.
•Therefore, it improves the network assortativity, however, as the network size increases, it becomes more
disassortative.
/
Fig. 31 Assortativity coefficient with different network
sizes and edge swap methods
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (9/16)
•Analysis of edge swap methods
•In Fig. 32, the average clustering coefficient of the networks against random ACCr, degree ACCdeg and
distance Adis dependent edge swaps, is calculated.
•The random and distance dependent edge swaps have not much effective
to make the cluster.
•It is because they do not consider the nodes degree during the optimization.
•Moreover, the degree dependent edge swaps connect similar degree nodes,
therefore, they have a high clustering coefficient.
/?
Fig. 32 Average clustering coefficient with different
Network size and edge swap methods
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (10/16)
•Analysis of edge swap methods
•In Fig. 33, the average shortest path length with the different number of nodes and random ASPLr, degree
ASPLdeg and distance ASPLdis dependent edge swap is shown.
•For all the edge swaps, the average shortest path length increases with
the increase in the number of nodes.
•The purpose of all the edge swap methods is to increase the network
robustness.
•However, the average shortest path length is compromised because in
the enhancement of network robustness the initial network structure is
compromised.
•So, from the results presented in Fig. 33, the conclusion can
be made that the robustness and average shortest path length are negatively correlated for SFNs.
/.
Fig. 33 Effects of edge swap on average shortest path length
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (11/16)
•Analysis of edge swap methods
•Fig. 34 shows for random GCEr, degree GCEdeg and distance GCEdis dependent edge swaps the global
communication efficiency is analyzed.
•It decreases as the number of nodes is increased.
•When the number of nodes is 100, the network has maximum global
communication efficiency.
•However, when the number of nodes is increased the shortest path
length is decreased.
•The decrease in the global communication efficiency is due to
presence of large number of nodes between the two specific nodes.
•The results prove that the edge swaps have no effects on global
communication efficiency.
//
Fig. 34 Effects of edge swap on global
communication efficiency of network
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (12/16)
•Analysis of edge swap methods
•Fig. 35 computational time for random Tr, degree Tdeg and distance Tdis dependent edge swaps is shown.
•The increase in the number of nodes causes the number of edges
to be increased.
•Therefore, the computational time to perform the edge swap also
increases.
•In the degree dependent edge swap, the computational time is
•high because the degree of all the nodes is calculated for each
edge swap.
•Moreover, the nodes that have independent edges are required to
calculate their respective degree information to make edges between similar degree nodes.
•However, the computational time of random and distance dependent edge swaps is low as compared to the
degree dependent edge swap.
•In random edge swap no extra information about the nodes is required. Furthermore, in the distance
dependent edge swap, only the information about the coordinates of the nodes is required.
/:
Fig. 35 Computational complexity of different edge
swaps
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (13/16)
•Analysis of edge swap methods
•In Fig. 36, the network robustness for the random Rr, degree Rdeg and distance Rdis dependent edge swaps is
shown.
•The initial network has low robustness as compared to the networks
obtained after the random, degree and distance dependent edge swaps.
•However, the random edge swap has a very small improvement in the
robustness as compared to the initial network.
•Although the degree dependent edge swap has high computational
time, however, it provides the highest robustness value as compared
to the distance dependent edge swap.
•Moreover, the degree dependent edge swap connects the nodes with similar degrees, therefore, a better
onionlike structure is formed, which is robust against the malicious attacks on SFNs.
•The improved value of network robustness obtained from the distance dependent edge swap proves that the
long links in the network have a vital role in defining the robustness of the network. The robustness value is
not changing in a gradual way that is due to the increase in the maximum degree nodes and the involvement
of randomness in network generation.
:
Fig. 36 Effects of edge swap on the robustness of
the network
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (14/16)
•Nodes' addition in MCS
•In Fig. 37, the performance of JAYA, ROSE and SA against the addition of the nodes in the MCS for the
network size of 100 is analyzed.
•The networks are fragmented into multiple subgraphs, after the
malicious attacks, therefore, against different optimizations techniques, the
robustness with and without the addition of nodes in the MCS is studied.
•Through the addition of nodes in the MCS, robustness of the
network is improved in all the techniques.
•After the high degree nodes removal, the network is fragmented into
multiple subgraphs.
•Then, the edges are added between the nodes to increase the MCS size.
•When 10 nodes are removed, robustness decreases very fast. However, with the addition of nodes in the
MCS, the network connectivity is improved.
•There is a small rise in the robustness, before it approaches to zero. It is due to the presence of a large
number of low degree nodes.
•The low degree nodes form a subgraph that has a large number of nodes.
:
Fig. 37 Robustness with the addition of nodes in MCS
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (15/16)
•Nodes' addition in MCS
•In Fig. 38, the network's performance against the bridge edge removal for JAYA, ROSE and SA is shown.
•The SFNs have a small number of nodes that have a high
degree, therefore, the probability of bridge edge existence is high.
•The network robustness is affected by the removal of nodes and
edges.
•The effects of edge removal on the network robustness are less as
compared to the removal of nodes, due to the complex network
structure.
•The removal of bridge edge decreases the robustness of all the techniques.
•Initially, JAYA outperformed ROSE and SA by maintaining the
network connectivity.
•After the removal of 40 edges, the sudden change in robustness shows that the network becomes bridge less.
•However, in ROSE and SA the network connectivity is maintained which shows that against the removal of
bridge edges the network structure is robust.
:
Fig. 38 Network connectivity with bridge edge
protection
Contributions (3/3)
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Simulation results (16/16)
•Network strength against different attacks strength
•In Fig. 39a, initially, the less attack strength with the removal of 5% nodes from the network is analyzed.
• All the techniques have a downward trend against the removal of nodes.
•The nodes that are removed from the network are randomly selected. The robustness of the JAYA optimized
network reduces faster as compared to ROSE and SA until 30 nodes are removed.
•After that, the robustness of the network approaches zero. However, for the SA and ROSE the network
robustness decreases gradually.
•Moreover, in Fig. 39b and Fig. 39c, the robustness of the networks against the attack strengths 2 and 3,
respectively, are studied.
:;
Fig. 39 Robustness of the network with (a) attack strength 1 (b) attack strength 2 (c) attack strength 3
Contributions (3/3)
,
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Conclusions (1/3)
•Proposed ISFNs technique enhanced the robustness of the scalefree network.
•The edge’ degree based and nodes’ distance based operation not change the degree distribution.
•No extra cost is required.
•The optimized topology remains scalefree.
•Proposed technique performs better for different numbers of nodes.
•The technique help to get the perfect onionlike structure.
•Robustness of ISFNs is high against the random and malicious attacks.
•Against variable attacks ISFNs outperforms the ROSE and SA.
:
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
•To achieve high robustness, the network topology plays an important role.
•NTES is proposed to study the importance of topology design to enhance the robustness of SFNs.
• The smallsized networks are more robust against attacks, therefore, the network is evolved by
dividing the sensor field into two parts.
•The network becomes robust because the nodes are removed in one part, however, the second part
of the network is still connected.
•The effect of attacks on the whole network is considered and the network is optimized by GA,
ABC and BFO.
•The experimental results prove that the network robustness is increased against attacks.
•Nodes are removed using the kcore decomposition.
•The high degree attack is more vulnerable to the network as compared to the corebased attack.
• The NTES is compared with two existing algorithms including BA model and HC.
Conclusions (2/3)
:
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Conclusions (3/3)
•By considering the importance of SFNs in realworld applications, firstly, the network is optimized by the
parameterless optimization algorithm JAYA.
•It is performed by creating the exclusive edges of the best and worst topologies in the population during
each iteration.
•All the topologies as part of the population are improved after performing the optimization.
•Moreover, by considering the importance of edge swaps in SFNs, they are classified into three categories.
•After performing different edge swaps, the effects of attacks and topological parameters are analyzed.
•The results prove that the degree dependent edge swaps are more effective than the random and distance
dependent edge swaps.
•In addition, after performing the distance dependent edge swap, the robustness is improved. It proves that
the existence of long links between the nodes plays an important role against the malicious attacks.
•Moreover, when malicious attacks happen, the network is fragmented into multiple subgraphs.
•Therefore, the nodes are added in the MCS to enhance the robustness of the network. Furthermore, in the
network, bridge edges are important for network connectivity.
•These edges are protected by performing the edge swap of the neighboring nodes. Moreover, the network is
analyzed against different attack strengths. The results prove that the higher attack strength is more
vulnerable to the network.
:?
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Future works
•We will implement the ISFNs scheme and NTES in different realworld networks.
•The effects of random and malicious attacks on these networks will study.
•The heterogeneous nodes will be considered.
•The network structure different from the onionlike will be made to enhance the robustness.
•Network degree distribution different from powerlaw will be used for SFNs.
:.
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Conference publications
•Usman, M., Javaid, N., Abbas, S. M., Javed, M. M., Waseem, M. A., & Owais, M. (2021, July).
A novel approach to network’s topology evolution and robustness optimization of scale free netw
orks.
In Conference on Complex, Intelligent, and Software Intensive Systems (pp. 214224). Springer,
Cham.
•Abbas, S. M., Javaid, N., Usman, M., Baig, S. M., Malik, A., & Rehman, A. U. (2021, July).
An Efficient Approach to Enhance the Robustness of ScaleFree Networks. In International
Conference on Innovative Mobile and Internet Services in Ubiquitous Computing (pp. 7686).
Springer, Cham.
•Usman, M., Javaid, N., Khalid, A., Nasser, N., & Imran, M. (2020, June).
Robustness Optimization of ScaleFree IoT Networks. In 2020 International Wireless
Communications and Mobile Computing (IWCMC) (pp. 22402244). IEEE.
:/
Optimization strategies to enhance the robustness of scalefree IoT networks against random and
malicious attacks (MS Thesis presented by Muhammad Usman)
Enhancing robustness of scalefree IoT networks against random and malicious attacks
MS Thesis defended by Muhammad Usman on 9th July 2021
Thank you very
much
for your attention
::