Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

Machine Learning-based Intrusion Detection System

Through WPA3 Protocol in Smart Contract System
Mohammad Sayduzzaman 1; Jarin Tasnim Tamanna1; Muaz Rahman2
Sadia Sazzad1 2
Department of EEE, National Institute of Textile
1
Department of CSE, National Institute of Textile Engineering and Research (NITER),
Engineering and Research (NITER), Constituent Institute of the University of Dhaka,
Constituent Institute of the University of Dhaka, Savar, Dhaka-1350
Savar, Dhaka-1350

Tawhidur Rahman 3
3
Digital Security & Digital Diplomacy, ICT Division,
Agargaon, Dhaka-1207

Abstract:- Nowadays, the Internet has become one of the in the operating system to access www properly. One
basic human needs of professionals. With the massive potential rule is that if the adapter has a WPA2-PSK (AES)
number of devices, reliability, and security will be crucial method, a separate driver must be installed for proper
in the coming ages. Routers are common to provide us connectivity. Otherwise, that will not work [3], [6]. Network
with the internet. These routers can be operated in security is crucial in any system run by smart contracts [7].
different modes. Some routers use the Wifi Security Traditional security measures, however, might not be enough
Protocol (WPA) or WPA2, and the Wifi Alliance to recognize and stop advanced cyber threats. The goal of this
introduced WPA3 on 25 June 2018. There are a lot of project is to create a strong Intrusion Detection System (IDS)
papers regarding Smart Contract (SC)–based IDS as well that operates within the WPA3 protocol by utilizing machine
as Machine Learning-based IDS. Very few discuss learning techniques. To find any security flaws, the IDS will
combining SC and ML-based IDS for different examine user behavior, network traffic patterns, and anomaly
authentication processes. In this paper, we will discuss detection. The system’s integration with smart contracts will
how combining SC and ML plays a vital role in enable it to react to threats on its own and improve the overall
authentication. Also, we play the role of embedded IDS security posture of decentralized networks using the WPA3
system so that existing vulnerabilities of the WPA2 and protocol [8].
WPA3 can be reduced to 99.62%.
Growing Complexity and Sophistication of
Keywords:- Machine Learning, Smart Contract, WPA3 Cyberthreats: Conventional intrusion detection systems may
Protocol, Router, Cybersecurity, Data Analysis. find it more difficult to identify and stop new security flaws
and attacks as a result of the growing sophistication and
I. INTRODUCTION complexity of cyber threats [9]. Through the analysis of
massive amounts of data and the identification of patterns
With the rapid development of technology, security is suggestive of harmful behavior, machine learning presents the
one of the biggest threats now. This security threat is to the possibility of improving detection capabilities. The necessity
information we have, process, and transmit [1], [2]. While of robust security method: Robust security methods are
transmitting any critical information we should think of its necessary for decentralized systems that run on blockchain
security and also the existing vulnerability of the medium we networks and are managed by smart contracts to guard against
use. As we are highly dependent on Wifi for using the www cyberattacks, illegal access, and data breaches. It is imperative
for any data transmission, we should know its vulnerability to guarantee the integrity and security of these systems in
and be aware of it. Wifi maintains the Institute of Electrical order to preserve trust and dependability in decentralized
and Electronics Engineers (IEEE) 802.11 standards. In case platforms and apps. Enhanced security for WPA3: Wi-Fi
of using public wifi, we should avoid any type of financial security has advanced significantly with the release of the Wi-
transmission as well as we should refrain from transmitting Fi Protected Access 3 (WPA3) protocol, which offers more
any confidential or private information. In Fig.1. the general robust encryption techniques and defense against a wider
concept of SCandML based IDS is given. Wifi alliances have range of threats than its predecessors. Wi-Fi networks’
introduced 8 different models of existing wifi [3], [4]. Wifi 8 security posture can be further improved by integrating
will be adopted within 2028. Fig. 2. [5] describes different machine learning-based intrusion detection with WPA3,
generations of wifi. Where * marked wifi 0, 1, 2, 3 are from adding more lines of protection against possible at-tackers
retroactive inference. This is why they do not exist in official [10], [11].
nomenclature, and Wifi 8 will be adopted by 2028.
Generations of wifi and their adapter driver must be installed

IJISRT24MAR2149 www.ijisrt.com 2926

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

Scalable and adaptive security solutions: Machine intrusion detection with smart con-tracts. This enables
learning algorithms have the ability to adapt and evolve over proactive threat identification and response without the need
time, continuously learning from new data and improving for human interaction. The main contribution of the paper is–
their detection capabilities. By leveraging machine learning
within the context of smart contract systems, organizations  To improve IDS’s overall performance, we have
can deploy scalable and adaptive security solutions capable of combined machine learning with smart contract
addressing evolving cyber threats in decentralized implementation and put out a novel methodology.
environments [12]. Security automation based on smart  Our methods will be put into practice, making WPA3
contracts: Operating on blockchain networks, smart contracts more dependable and secure.
allow predefined rules and policies to be executed  We assess various models and combine the findings for
autonomously. Security measures can be automated and additional study.
enforced in real time by integrating machine learning-based

Fig 1 Concept of SCandML based IDS

 Organization: II. BACKGROUND AND LITERATURE REVIEW

This article is organized as follows: The author discusses
the previous works and basic idea of IDS in section II, along From the very beginning of wifi technology, Intrusion
with attack vector detection and their solution. Then, proposed Detection Systems played a vital role in detecting any
SC & ML-based IDS for WPA3 and Algorithm in section III. inhuman as well as unwanted activities that may breach the
The evaluation of the results and an adequate discussion are overall security and open the ground for critical cyber-attacks
presented in section IV. Finally, the conclusion, along with a [13]. IDS can be an anomaly or signature-based [14].
few considerations, limitations, and future discussions, is in Researchers are researching smart contract-based IDS as well
section V. as ML-based IDS. Recently, they have been combining
different technologies like SC & ML for better performance.
Below, we will discuss the adoption process of SC & ML,
followed by a section with previous work. Required
abbreviations are listed in Table 1.

Table 1 List of Common Abbreviations with Description

Keys Description
AI Artificial Intelligence
AFD Armed Forces Division
AP Access Point
Wifi Wireless Fidelity
www World Wide Web
WPA Wi-Fi Protected Access
WPA2 Wi-Fi Protected Access Version 2
WPA3 Wi-Fi Protected Access® 3
WEP Wired Equivalent Privacy
IEEE Institute of Electrical and Electronics Engineers
PSK Pre-Shared Key
AES Advanced Encryption Standard
SAE Simultaneous Authentication of Equals
SAE-PK Simultaneous Authentication of Equals Public Key
Sec Hash Extension Security Parameter

IJISRT24MAR2149 www.ijisrt.com 2927

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

SSID Service Set Identifier

AES-GCM AES With Galois/Counter Mode
EAP Extensible Authentication Protocol
RC4 Rivest Cipher 4
SSID Service Set Identifier
PAP Password Authentication Protocol
PEAP Protected Extensible Authentication Protocol
MS-CHAPv3 Microsoft Challenge Handshake Authentication Protocol Version 3
TTLS Tunneled Transport Layer Security
ECDSA Elliptic Curve Digital Signature Algorithm
IDS Intrusion Detection System
SC Smart Contract
ML Machine Learning
MFP Management Frame Protection
EAP Extensible Authentication Protocol
KCM Key Confirmation Messages
DApp Decentralized Application or System
SSI Self-Sovereign Identity
ISI Information Sciences Institute
USC University of Southern California

While researching from the very beginning, we found found that Random Forest detects 99.62% of unwanted
huge papers full of hesitation about whether WPA3 is the authorization approaches that fail to connect via WPA3.
miracle of technology that can never be backdated. [15] or it
is just a simple improvement over WPA3, which is still Normally, wi-fi technology exchanges information via
vulnerable to all those attacks that are very common in hand-shake. In the handshaking device the router shares a
WPA2, while WPA3 is configured to serve both WPA2 & hashed version of the Wi -Fi password. One major flaw is that
WPA3 supported devices [16]. We approach further to find an attacker within range can simply listen in and capture the
something exclusive, and then we find several ML approaches hash password when you connect. Once the handshake is
for securing WPA3, but few of them are standard as we have captured, they can just leave and use a program like Hashcat
to maintain an easy-to-implement approach for WLAN or to try to crack the network password. Now, if the attacker
Wifi [17]. Next, we go for technology like blockchain; we isn’t in range when a device is connecting, they don’t really
found and interesting part of Blockchain is Ethereum-based need to wait. They can just force a handshake by sending
smart contracts. Both the server and the client will management frames called de-auth packets. This will
communicate, and no third party will be involved; the disconnect the actual device from the Wi-Fi network, forcing
transaction will be recorded on a universal ledger. Then, for it to exchange a new handshake anyway while the attacker
authentication and intrusion detection, we take ML [18]. We just listens in. This is the second major flaw. Table II shows
have tried different ML algorithms so that different vendors the difference between WEP and different WPA versions
can choose suitable options for them. In our system, we have [19], [20].

Fig 2 Overviews of Timeline for Wi-fi Network

IJISRT24MAR2149 www.ijisrt.com 2928

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

Table 2 Difference between Different WPA Versions

Type WEP WPA WPA2 WPA3
Encryption Method RC4 RC4/TKIP AES AES-256 in GCM mode
with SHA-384
Encryption key size 40 bits 128 bits 128 bits 128 bits, 192 bits, and 256
bits
IV size 24 bits 48 bits 128 bits 256 bits
Authentication process Weak 802.1x -EAP PEAP-MSCHAPv3, EAP- SAE, EAP
TTLS/PAP, and EAP-TLS
Data Integrity CRC 32 - ICV MIC CCMP AES-GCM

From Table 2, we can see that WEP uses RC4 as an jamming a WPA3 network by forging De-Auth packets
encryption method, where each encryption key size is about between a router and any connected devices.
40 bits, and the initialization vector is 24 bits, where data
integrity is CRC 32- TCV, but its authentication process WPA3 also includes a feature called forward secrecy,
proved weak. WPA3 comes with huge coverage. It has the and what this means is that if someone was able to capture
following key features: some of your Wi-Fi traffic and later on learned your
password, they wouldn’t be able to go back and decrypt
 WPA3-Personal Only Mode. everything that they gathered. Now for this and all the other
 WPA3-Personal Transition Mode. security updates, you should definitely update to WPA3.
 WPA3-Personal SAE-PK only Mode. could leak data, the creators of WPA3 were forced to react
 WPA3-Enterprise only Mode. with a hasty fix to address the timing attack, but in doing so,
 WPA3-Enterprise Transition Mode. they also made it possible to jam WPA3. So, to break down
 WPA3-Enterprise 192-bit Mode. the original issue, when a device joins a WPA3 network, the
router converts, while, unfortunately, a nearby attacker can
WPA3 adopted the Management Frame Protection measure this by the time it takes for the router to reply, and
(MFP) and now WPA3 is harder to brute-force or dictionary this information lets the attacker more easily brute force the
attack. In WPA3, credentials comprise of [21]. WiFi password. Then, based on the unique amount of time it
takes for WPA3 routers to process different passwords,
 Here is the Fingerprint Equation for WPA3: hackers can rule out large groups of password guesses to try
brute-forcing attacks. This breaks WPA3’s promise of
 Fingerprint =L(Hash(SSID||M||K_ AP ), 0, 8 ∗ Sec + immunity from brute-forcing attacks less than a year after its
release. They decide to solve this by always making the
19∗λ/4−5)
access points perform a lot of computations and always reply
a bit slower. The way that we made it reply slower is by
 And the Password shall then be Determined as Follows:
making this algorithm that is used internally perform a few
iterations of a certain function. Now, this prevents the timing
 Password Base=Base32(P (0)||P (1)||. . . ||P (λ/4− 1))
leak if done properly. However, doing these iterations adds a
 Password=Add Separators (P asswordBase||ChkSum) lot of overhead. So if you then, for example, implement WK3
on a very lightweight device or an IoT device, this
More details of this equation is available for further countermeasure means they possibly are vulnerable to denial-
research at https://www.wi-fi.org/system/files/WPA3 of-service attacks, or they don’t or implement just a weak
Specification v3.3.pdf. In WPA3, a Dragonfly handshake is version and then they might be vulnerable to the side
designed to frustrate attackers. While a nearby attacker could channels. As mentioned, their fix paved the way for a fairly
capture a handshake, it takes too long to brute force, and they simple denial-of-service attack. Hackers can send many
have to stay in range of the router to try every single password handshakes at the same time, causing the router to crash and
guess. De-Auth packets and other management frames are taking the network completely offline [22], [23].
also now encrypted in WPA3, which prevents attackers from

IJISRT24MAR2149 www.ijisrt.com 2929

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

Fig 3 (a) 4-Way Handshake (b) Shared Key Authentication Process. (c) Existing WPA3 Model and
(d) A Proposed Model for WPA3.

In Fig. 3. among these four scenarios in case (a) after and case (d) is the proposed model. ML will monitor, and SC
authentication and association, a 4-way handshake takes place will make an entry. After a successful verification process, it
using Extensible Authentication Protocol. This represents an will improve further. Table III shows a list of existing works
open system authentication. Case (b) represents a shared key that were analyzed on WPA3 IDS.
authentication process. Case (c) is the existing WPA3 model,

IJISRT24MAR2149 www.ijisrt.com 2930

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

Table 3 List of Attacks and Necessary Tools Along with their Solutions
Attacks Attack method Impact Mitigation Process
Downgrade to serve WPA2 devices Client connects to AP becomes vulnerable Device firmware to be updated
WPA2 rather WPA3
Authentication Flood attack Denial of service AP goes offline SC & ML based IDS
SAE out of range attack Denial of service AP goes offline SC based real time token
SAE unsupported group attack Denial of service AP goes offline SC based group token
Downgrade group attack AP flooding AP becomes vulnerable ML based authentication
Timing side channel attack Brute-force System leaks information SC & ML based IDS
about the password
Deauthentication Attack Brute-force / dictionary AP goes offline SC & ML based IDS
Beacon/Probe flood attack Response flooding Confuse clients find the SC & ML based IDS
legitimate AP

Table 3 shows a list of attacks and necessary tools along with their solutions [16], [24]. Previously, some commercial
signature-based IDS was developed after the failure of anomaly-based IDS [25] likely Snort-Wireless [26], AirMagnet [27] and
AirDefence [27] were successful for WPA2. However, they are not updated enough to work successfully for WPA3. Here comes
the concept of SC & ML-based IDS systems. In Table IV a list of previous work is given in a brief.

Fig 4 Proposed Architecture for SCML-based IDS for WPA3 Network

Table 4 List of Existing Works Analysis on WPA3 IDS:

Work Platform Contribution Architecture Method
Christopher P. Kohlios et.al [26] Four-way handshake Attack vector WPA2, WPA3 Attack flow analysis
attack flow. analysis
Dalal, Neil, et al. [27] Attack vector Signature based IDS WPA2, WPA3 Signature based IDS
Saini, Rahul et al. [18] WPA3 enterprise IDS and Attack flow WPA3 Signature based IDS
network Protected
Alsharbaty, Firas S an Ali et al. [28] Protected wifi Electronic substation WPA3 Hybrid IDS
Bhutta, A.A et al. [29] LightGBM Realtime WPA3 Lightweight wifi IDS
Koutras, Dimitris et al. [30] Automatated IDS Automation WPA3 ID based method
Thankappan, M., Rifa Pous et al. [31] Protected wifi Signature based IDS WPA3 Signature based IDS
Saifan, Ramzi and Radi et al. [32] Log Monitoring Mitigationtool WPA2 Log Monitoring
Kishiyama , Brian and Guerrero et al. Security Policies Security Policies WPA2 Security Policies
[33] Automation Automation Automation
Qaddoori, Sahar L and Ali et al. [34] Industrial IOT IDS WPA2 ML
Uszko,Krzysztof and Kasprzyk et al. 5G WAN Rule based IDS WPA2 ML
[35]
Narayana, Datta Sai et al. [36] Protected Network Prevent Hijaking WEP Adaptive threat
defebding
Stella,Kand Menaka et al. [37] Wifi Hotspot Traffic monitoring Detecting Threat defebdibg
spoofing
Mansour, Salah Eddine et al. [38] IoT-Fog Networks Improving security IOT AI Image processing

IJISRT24MAR2149 www.ijisrt.com 2931

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

III. PROPOSED SCML-BASED IDS FOR WPA3

Fig. 4. is the Proposed Architecture for SCML-based IDS for WPA3. Fig. 5. Describes the step-by-step method for our
proposed SCandML-based IDS system in (Fig. 4.):

Fig 5 Proposed Methodology of SCandML based IDS

 Data Collection:  Real-Time Monitoring and Analysis:

Obtain network traffic data from Wi-Fi access points Use the integrated solution to deploy real-time network
that are using the WPA3 protocol, such as packet headers, traffic monitoring. Use the machine learning models that have
payload details, and metadata. Gather extra contextual data on been trained to analyze incoming data streams and find any
user activity, device information, and network topology [39], irregularities or potential security issues [46].
[40].
 Automated Response Mechanisms:
 Feature Extraction and Labeling: Incorporate automated reaction systems inside the smart
To depict network traffic patterns and behavior, extract contracts to address identified security risks. Removing access
pertinent elements from the data that has been gathered. privileges, upgrading encryption keys, barring questionable
Packet size, connection frequency, protocol utilization, source people, and setting off alarms for additional research are a few
and destination IP addresses, and timestamps are a few examples of possible actions.
examples of features. Moreover, the gathered data will be
classified according to predetermined criteria as either normal  Continuous Learning and Adaptation:
or abnormal [41], [42]. Retrain and up-date the machine learning models
frequently in response to input from ongoing network research
 Smart Contract Governance: and monitoring. Include systems that allow response plans
Create smart contracts to control how the decentralized and detection levels to be dynamically adjusted in order to
network’s intrusion detection system operates. Establish respond to changing security threats.
guidelines and procedures, such as thresholds for anomaly
detection and the steps to be taken after discovery to identify
and address security threats [45].

IJISRT24MAR2149 www.ijisrt.com 2932

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

 Evaluation and Optimization:  Smart Contract Design:

Assess the intrusion detection system’s performance on Parameters for creating smart con-tracts that regulate the
a regular basis with regard to response efficacy, false intrusion detection process, including as access control
positive/negative rates, and detection accuracy. To increase methods, situations that set off alerts or responses, and
the system’s effectiveness over time, optimize it based on contract logic.
performance indicators and input from security incidents.

By using this approach, businesses can leverage the

powers of machine learning, the WPA3 protocol, and smart
contract governance to create a strong Machine Learning-
based Intrusion Detection System through WPA3 Protocol in
Smart Contract System that improves security within
decentralized networks. We have demonstrated how our
suggested method-ology can be employed with the current
WPA3 design in Fig. 3.

 ML-based Intrusion Detection Process

Within the field of artificial intelligence (AI), machine
learning is a subset that allows systems to learn from data and
make judgments or predictions without requiring explicit
programming. To find patterns and anomalies suggestive of
security dangers, machine learning algorithms are trained on
massive databases of historical security-related data in the
field of cybersecurity. Among the crucial facets of machine
learning are:

 Training Data
 Feature Extraction
 Model Training
 Model Evaluation and Validation, and Fig 6 Key Parameters of SCandML based IDS
 Deployment and Monitoring
 Blockchain Integration:
A number of parameters are crucial to the design, Issues including data storage, off-chain data access,
implementation, and assessment of an intrusion detection transaction fees, and blockchain scalability considerations that
process based on machine learning and smart contracts. Fig. 6 are involved in combining machine learning models with
represents some crucial factors: smart contracts on a blockchain network [47].

 Data Sources:  Real-Time Monitoring and Detection:

Parameters pertaining to the data sources—such as Specifications such as sample rates, data streaming
system logs, network traffic logs, sensor data, or external protocols, processing delay, and resource requirements for
threat intelligence feeds—that are utilized to detect intrusions. continuous monitoring that are connected to real-time network
activity monitoring.
 Feature Selection:
Factors, such as feature types, feature extraction  Automated Response Mechanisms:
strategies, and feature engineering techniques, that are In smart contracts, automated reaction mechanisms can
involved in choosing pertinent features from the data to train be specified. These include response actions, thresholds for
machine learning models. starting reactions, and escalation processes for handling
security incidents.
 Machine Learning Algorithms:
Parameters pertaining to the selection of machine  Continuous Learning and Adaptation:
learning algorithms for intrusion detection, including Details include update frequency, retraining methods,
unsupervised anomaly detection techniques, decision trees, feedback systems, concepts, and data drift management
random forests, support vector machines, and neural procedures that are related to the intrusion detection system’s
networks. continuous learning and adaptability.

 Model Training:  Performance Metrics:

Hyperparameters, optimization methods, cross- For evaluating the recall, accuracy, precision, F1-score,
validation techniques, and training/validation/testing data false positive and false negative rates, detection latency, and
splits are some of the parameters that are involved in the scalability of the intrusion detection system can be
training of machine learning models. considered.

IJISRT24MAR2149 www.ijisrt.com 2933

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

 Community Consensus and Governance: Authentication Service This component verifies the
Characteristics of the decentralized network’s user’s identity and credentials. It may involve traditional
governance and community involvement, including voting methods like username/password, bio metric authentication,
procedures, stakeholders’ involvement, consensus algorithms, or decentralized identity solutions such as Self-Sovereign
and transparency standards, are part of Community Consensus Identity (SSI).
and Governance. Behavioral analysis and machine learning
approaches are frequently used in cybersecurity to improve Smart contracts contain the authentication logic for the
threat detection capabilities, which helps to identify and DApp. They define the rules and criteria for authenticating
address security issues more successfully [48]. Through the users and granting access to the application’s features and
early detection of suspicious activity and possible security functionalities.
breaches, these strategies assist to stay ahead of cyber threats.
In some situations, such as decentralized applications or Blockchain Network The smart contract(s) run on a
systems where access control is controlled via a blockchain blockchain network, which serves as the underlying
network, smart contracts can make barring a user easier. infrastructure for the DApp. The blockchain ensures the
integrity and security of user authentication processes.
 To Block a user, a Smart Contract can be Created as
Follows: Consensus Protocol The consensus protocol governs
User Identification The smart contract needs a way to how transactions and smart contract state changes are
identify the user or account that should be blocked. This could validated and agreed upon by network participants. It ensures
involve the user providing some form of unique identifier, the immutability and trustworthiness of the blockchain
such as an Ethereum address or a digital identity stored on the network.
blockchain.

Blocking Mechanism, the blocking mechanism’s

implementation is handled by logic in the smart contract. This
rationale outlines the circumstances in which a user should be
blocked as well as the steps to be performed when blocking
has been decided.

Authorization and Access Control The smart contract

checks users’ credentials and permissions before enabling
them to engage with the system or application. A user is
blocked by the smart contract if their access requests are
rejected in the future, depending on pre-established rules or
criteria.

Event Logging and Notification The smart contract may

log the action on the blockchain or emit an event when a user
is blocked. In addition to promoting openness, this makes the
blocking event visible to other network users.

Revocation of Privileges The smart contract may remove

some of the prohibited user’s rights or privileges, depending
on the application or system requirements. In a decentralized
social media site, for instance, a blocked user might no longer
be able to communicate with other users or upload material.

Appeal or Dispute Resolution In certain situations, the

smart contract might have procedures that allow users to con-
test actions taken against them or appeal-blocking decisions.
This can entail giving people a way to submit proof or contest
the legitimacy of the blocking event.

 Approaches of Smart Contracts

Fig. 7. illustrated the flow diagram for controlling user
access via smart contract:

User Interface The user interface is where users initiate

the authentication process. This could be a web interface,
mobile app, or desktop application. Fig 7 Flow Diagram for Controlling user Access via
Smart Contract

IJISRT24MAR2149 www.ijisrt.com 2934

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

WPA3 Handshaking Process Fig. 3(c). explained an requirements by supplying the session key or other credentials
overview of existing WPA3 handshaking process, which is needed for WPA3 authentication.
discussed below [16]:
 Secure Communication:
SAE Handshake Initialization: The SAE handshake After authorization, the user’s device can use the
begins with the client sending a ”commit” message containing standard WPA3 protocols to create a secure connection with
some unique IDs like MAC address. The access point (AP) the access point. According to WPA3 standards, all
responds with its own commitment value and generates a communication between the user’s device and the access point
random nonce. Both parties use their commitment values and is encrypted and safe [50].
the nonces to calculate the shared secret.
 Transparency:
 Step 1: The client and AP deal with the Pairwise Master All-access control transactions and inter-actions with
Key (PMK). The PMK is then used to generate other keys smart contracts are recorded in an immutable ledger that
such as the Pairwise Transient Key (PTK) and Group blockchain technology offers. Network administrators are able
Transient Key (GTK) which are prepossessing tasks to to detect and monitor user access to Wi-Fi networks in real-
establish a connection. time, which also provides transparency.
 Step 2: Exchanging messages is important to confirm the
position of the secret key. This includes the exchange of  Decentralized Network Governance:
Key Confirmation Messages (KCMs) to verify the Decentralized management of Wi-Fi network
integrity of the key exchange process. configurations and policies is another application for smart
 Step 3: Once verification and key exchange process is contracts. Smart contracts might regulate network factors like
completed then the secure connection is established. In key management, access controls, and encryption techniques,
this process negotiated encryption algorithm (e.g., GCMP- offering a visible and im-penetrable governance structure
256) can be used. [51]. However, when putting such solutions into practice in
 Step 4: By changing the encryption keys on a regular real-world circumstances, it’s crucial to take into account
basis, WPA3 maintains forward secrecy, making it more practical issues like scalability, interoperability, and
difficult for an attacker to decrypt previous regulatory compliance. Integrating smart contracts with Wi-Fi
conversations—even if they are able to access the keys. Protected Access 3 (WPA3) in the event of a hacking attack
This is a considerable improvement in WPA3 over WPA2. involves leveraging blockchain technology’s inherent security
features to mitigate and respond to such attacks [52].
 The Integration of Smart Contracts
The integration of Smart contracts with Wi-Fi Protected  Here’s how this Integration could Enhance Security and
Access 3 (WPA3) would include blockchain technology with Resilience:
WPA3’s permission and authentication protocols. Here is a
theoretical explanation of how this integration might function:  Real-Time Threat Detection:
Smart contracts have the ability to continuously scan the
Decentralization of Identity Management: Manage user blockchain network for any questionable WiFi-related
IDs for Wi-Fi access by utilizing blockchain-based activities. The smart contract has the ability to send out alerts
decentralized identity systems like Self-Sovereign Identity in response to anomalies like repeated unsuccessful
(SSI). Every user might have a distinct digital identity that is authentication attempts or unwanted access attempts.
kept on the blockchain and contains authorization codes and
login credentials for using Wi-Fi networks.  Automated Response Mechanisms:
It is possible to program smart contracts to launch
 Using Smart Contracts for Access Control automated reactions in response to attempted hacking. For
On a blockchain network (like Ethereum), smart instance, the smart contract might dynamically modify access
contracts can be used to control access to WiFi networks. The control settings to improve security or momentarily stop
access point can communicate with the blockchain network additional access attempts from the shady source in the event
through smart contracts when a user tries to connect to a that it detects a brute-force attack.
WPA3-secured Wi-Fi network in order to confirm the user’s
identity and authorization [49].  Secure Recovery Processes:
After a hacker assault, smart contracts can enforce safe
 Authentication and Authorization: recovery procedures like reissuing login credentials or
The user’s device can start a transaction with the access resetting access rights. With the use of smart contracts, these
control smart contract once it is connected to the Wi-Fi recovery procedures may be regulated and automated,
network. Based on specified rules stored on the blockchain, lowering the possibility of additional security breaches or
the smart contract can confirm the user’s digital identity and human error [53].
determine whether they have the required permissions to
access the network. The smart contract may authorize access
if the user satisfies the authentication and authorization

IJISRT24MAR2149 www.ijisrt.com 2935

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

 Community Consensus Mechanisms:  Dynamic Access Control Policies:

When it comes to Wi-Fi security, smart contracts have Within the smart contract, define dynamic access control
the potential to im-prove network resilience by utilizing restrictions depending on contextual data such device kind,
community consensus to evaluate and address new threats location, time of day, and user behavior. Real-time access
jointly. Organizations can adopt a more proactive and resilient privilege adjustments can be made for suspicious users, either
approach to cybersecurity by combining smart contracts with by temporarily removing their access or reducing it.
WPA3 and blockchain technology. By utilizing automated
threat detection, response mechanisms, and immutable audit  Rate Limiting and Throttling:
trails, these strategies can significantly lessen the impact of Use rate-limiting and throttling techniques in the smart
hacking attacks. contract to restrict the quantity of connections or requests that
a single user may make in a certain amount of time. If a user
 Integration with Machine Learning IDS exceeds the permitted criteria, they may be automatically
Smart contract data can be analyzed by machine learning blocked.
algorithms, which can then be trained to spot trends that point
to possible security risks or questionable activity. Anomalies  Behavioral Analysis and Machine Learning:
such as strange connection attempts, strange traffic patterns, Make ad-vantage of machine learning techniques to
or recognized attack signatures can be identified by the examine trends in user behavior and spot deviations that point
machine learning model [54]. to questionable activities.

 Smart Contract-Based Decision Making: To increase the machine learning model’s accuracy in
The machine learning model has the ability to use the identifying malicious activity, train it with historical data. In
smart contract to send out alerts or notifications when it this paper, we will focus on Machine Learning and Behavioral
notices questionable activities. Based on predetermined rules Analysis [55].
or regulations, smart contracts are able to assess the context
and degree of detected anomalies and decide on the best  Behavioral Analysis
course of action. In behavioral analysis, entities like individuals, devices,
apps, or networks are observed, and their behaviors,
 Automated Response and Mitigation: interactions, and patterns of behavior are analyzed. The aim is
To reduce any security risks, smart contracts can start to find abnormalities in behavior that might point to
automated reaction systems. This could entail limiting access, malevolent or illegal activities. Among the crucial facets of
barring questionable people for a short while, or changing behavioral analysis are:
security settings like encryption keys.
 Baseline Establishment
 Continuous Learning and Adaptation:  Anomaly Detection
The machine learning model is able to adjust its  Contextual Establishment, and
detection capabilities to threats that change over time by  Continuous monitoring
continuously learning from new data. Over time, smart
contracts can help to increase the efficacy of the IDS by  Description of Proposed Algorithm
facilitating the integration of updated security policies and In a nutshell, the process can be represented in an
machine learning models. Here are some techniques that can algorithm. The proposed architecture firstly collects a flag
be implemented within smart contracts to block suspicious value which is uniquely assigned from the system to the user
users in the context of Wi-Fi Protected Access 3 (WPA3): who wants to access the system to use the data. Then, the
value, along with the information, was passed through the
 Threshold-based Anomaly Detection: smart contact. In the smart contract system, the value is
Set thresholds in the smart contract to track different updated from the flag value to another unique value where a
data, such as the quantity of unsuccessful authentication private key is added. Then, the smart contract checks whether
attempts, odd traffic patterns, or abrupt activity spikes. there is any anomaly directed or not, comparing the value
Consider the user to be suspect and start blocking mechanisms with the system. If the anomaly is detected, the value is then
if the detected metrics are higher than predetermined levels. returned to the smart contract, but if there is no anomaly, then
the value is passed to the machine learning model. Again, the
 Whitelist and Blacklist Management: value is updated, and the value is with the system and
Keep lists of recognized, trustworthy, and questionable authenticates the user as a valid user if the value is matched.
users on both the whitelist and the blacklist within the smart As a result, the invalid user can be directed, and the data
contract. Put the user on the blacklist when suspicious activity sharing security is ensured.
is discovered to stop them from accessing the network again.

IJISRT24MAR2149 www.ijisrt.com 2936

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

of cases that, out of all actual threats, are mistakenly

categorized as non-threats.

 Detection Capabilities:
Examine the system’s capacity to identify various
security risks, such as known assaults, zero-day exploits, and
unusual activity. Evaluate how sensitive the system is to
variations in the threat landscape, attack patterns, and network
circumstances.

 Response Effectiveness:
Analyze how well smart contract-triggered automated
reaction mechanisms mitigate identified security threats.
Evaluate how quickly and suitably response actions—like
stopping malicious activity, removing access, or initiating
incident response workflows—are carried out.

 False Positive Analysis:

Examine cases of false positives to find the root reasons,
which may include noisy data, incorrectly constructed models
or inadequate feature selection. To reduce false positives
without sacrificing detection accuracy, change the response
thresholds, feature selection criteria, or model parameters.

 False Negative Analysis:

Analyze false negative cases to comprehend security
dangers that were overlooked and possible intrusion detection
system blind spots. Improve feature engineering and model
training to increase sensitivity to new or subtle security
threats.

 Feedback and Iterative Improvement:

In summary, the DApp authentication process involves Collect feedback regarding system performance,
the user interface initiating authentication, an authentication response actions, and secu-rity incidents in order to
service verifying user identity, smart contracts containing continuously enhance the intrusion detection process. Apply
authentication logic, and a blockchain network ensuring the the knowledge gained to feature selection, smart contract
security and integrity of the authentication process. governance, model retraining, and system architecture as a
whole.
IV. RESULT ANALYSIS
 Scalability and Resource Efficiency:
A machine learning and smart contract-based intrusion Examine the intrusion detection system’s scalability to
detection process’s performance, efficacy, and efficiency in manage rising network traffic volumes and evolving security
identifying and averting security risks within a decentralized threat complexity. Optimize the use of resources, including
network are assessed as part of its result analysis. Key com- memory, bandwidth, and compute power, to guarantee the
ponents of result analysis include the following: system operates effectively in decentralized contexts.

A. Performance Metrics  Overall System Reliability and Trustworthiness:

Assess the intrusion detection system’s dependability
 Accuracy: and credibility in preserving the security and integrity of the
The percentage of security threats that are accurately network. To determine the overall effectiveness of a system,
identified out of all instances. Precision: The percentage of measure user satisfaction, system uptime, and adherence to
security threats successfully recognized out of all instances security standards. Organizations can improve the security
that are categorized as threats. Recall: The percentage of posture of decentralized networks by doing thorough result
security risks that were accurately recognized out of all real analysis to pinpoint areas for improvement and strengths and
threats. F1-score: A fair performance indicator derived from weaknesses in their Machine Learning and Smart Contract-
the harmonic mean of recall and precision. False Positive based Intrusion Detection Process.
Rate: The percentage of cases mislabeled as threats out of all
cases that are not threats. False Negative Rate: The percentage

IJISRT24MAR2149 www.ijisrt.com 2937

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

Fig 8 Confusion Matrix of Different Methods

B. Dataset  Class Imbalance:

A popular benchmark dataset in the fields of network A problem with the KDD Cup’99 dataset is its class
security and intrusion detection is KDD Cup’99 [56]. It was imbalance, with most cases being associated with legitimate
made in 1999 specifically for the Third International activity and only a tiny proportion being malicious attacks.
Knowledge Discovery and Data Mining Tools Competition. This imbalance needs to be handled carefully during model
The dataset was developed by researchers from the training and evaluation since it can impact how well machine
Information Security Institute (ISI) at the University of learning models function.
Southern California (USC) and the Information Sciences
Institute (ISI), along with other contributors. We also use this  Preprocessing Requirements:
dataset for our system. Some key features of the KDD Cup’99 Preprocessing the dataset is frequently necessary before
dataset: using it for machine learning studies because of its size and
complexity. This could involve fixing class imbalance,
 Purpose: encoding category data, handling missing values, and
The purpose of the dataset creation was to support normalizing numerical features [57].
research in the areas of network security and intrusion
detection. It is made up of numerous examples of network  Use in Research and Benchmarking:
traffic data, including both benign and malevolent activity. Researchers and practitioners have utilized the KDD
Cup’99 dataset ex-tensively to evaluate intrusion detection
 Data Characteristics: systems, assess machine learning algorithm performance, and
A variety of network traffic statis-tics, including both compare various methods for identifying and thwarting
common and malicious activity, are included in the dataset. network attacks.
Numerous network protocols are covered, including TCP,
UDP, ICMP, and others. Because the data instances are All things considered, the KDD Cup’99 dataset
labeled with distinct attack kinds, supervised learning tasks continues to be an important tool for furthering research in
can benefit from using them. intrusion detection and network security, offering a consistent
bench-mark for assessing the efficacy of different detection
 Data Format: methods and algorithms. Fig. 8. Describes the Confusion
The dataset is presented in tabular style, where a Matrix for different ML models were,
network connection or session is represented by each row.
Numerous attributes or features, such as the protocol type,  Confusion Matrix for KNN
service type, source and destination IP addresses, etc.,  Confusion Matrix for KStar
describe the properties of the network connection in each row.  Confusion Matrix for Logistic Regression
 Confusion Matrix for Na¨ıve Bayes
 Confusion Matrix for Random Forest
 Confusion Matrix for SGD

IJISRT24MAR2149 www.ijisrt.com 2938

Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

We have used total 25193 data and among them V. CONCLUSION

performance testing was done on 80percent data and Training
data was 20 percent. Our outcomes are as follows: Organizations can effectively and accurately identify a
variety of security concerns, such as well-known assaults and
Fig. 9. Is the comparison chart of the output of different unusual patterns, by utilizing a combination of machine
methods we have used. Where (a) Calculates the accuracy of learning algorithms. A decentralized governance framework
different ML models. and (b) Calculates the accuracy against for automating threat detection, initiating actions to identified
Time. By observing Fig. 9. and Fig. 10. whether or not time is threats, and guaranteeing accountability and transparency in
taken into account, Random Forest stands out among the rest. security operations is offered via smart contracts. The
amalgamation of machine learning with smart contracts
C. Accuracy Analysis with respect to Time provides decentralized networks with enhanced security
posture through automatic reaction capabilities, adaptive
learning, and real-time monitoring. In the future, Advanced
Machine Learning Techniques: Subsequent investigations
may examine the application of sophisticated machine
learning methodologies, such deep learning and reinforcement
learning, to enhance the precision and efficiency of intrusion
detection systems. Overall, the field of machine learning and
intrusion detection based on smart contracts has enormous
potential to transform cybersecurity in decentralized systems
in the future. Through the consideration of the previously
described factors and the adoption of continuous
technological progress and cooperation, institutions can
construct robust and flexible security systems that are able to
repel dynamic cyberattacks in the decentralized environment.

REFERENCES
Fig 9 Comparison Chart of Different Method we have used
[1]. A. Rahman, M. J. Islam, Z. Rahman, M. M. Reza, A.
Anwar, M. P. Mahmud, M. K. Nasir, and R. M. Noor,
“Distb-condo: Distributed blockchain-based iot-sdn
model for smart condominium,” IEEE Access, vol. 8,
pp. 209 594–209 609, 2020.
[2]. A. Rahman, M. J. Islam, A. Montieri, M. K. Nasir, M.
M. Reza, S. S. Band, A. Pescape, M. Hasan, M.
Sookhak, and A. Mosavi, “Smartblock-sdn: An
optimized blockchain-sdn framework for resource
management in iot,” IEEE Access, vol. 9, pp. 28 361–
28 376, 2021.
[3]. E. Oughton, G. Geraci, M. Polese, and V. Shah,
“Prospective evaluation of next generation wireless
broadband technologies: 6g versus wi-fi 7/8,”
Available at SSRN 4528119, 2023.
[4]. A. Rahman, M. K. Nasir, Z. Rahman, A. Mosavi, S.
Shahab, and B. Minaei-Bidgoli, “Distblockbuilding: A
distributed blockchain-based sdn-iot network for smart
building management,” IEEE Access, vol. 8, pp. 140
008–140 018, 2020.
[5]. L. G. Giordano, G. Geraci, M. Carrascosa, and B.
Bellalta, “What will wi-fi 8 be? a primer on ieee
802.11 bn ultra high reliability,” arXiv preprint
arXiv:2303.10442, 2023.
[6]. M. T. Ahmed, R. Islam, M. A. Rahman, M. J. Islam,
A. Rahman, and Kabir, “An image-based digital
forensic investigation framework for crime analysis,”
in 2023 International Conference on Next-Generation
Computing, IoT and Machine Learning (NCIM).
IEEE, 2023, pp. 1–6.
Fig 10 Accuracy of ML Models

IJISRT24MAR2149 www.ijisrt.com 2939

 Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

[7]. M. J. Islam, A. Rahman, S. Kabir, M. R. Karim, U. K. [19]. K. A. Adbeib, “Comprehensive study on wi-fi security
Acharjee, M. K. Nasir, S. S. Band, M. Sookhak, and S. protocols by analyzing wep, wpa, and wpa2,” African
Wu, “Blockchain-sdn-based energy-aware and Journal of Advanced Pure and Applied Sciences
distributed secure architecture for iot in smart cities,” (AJAPAS), pp. 385–402, 2023.
IEEE Internet of Things Journal, vol. 9, no. 5, pp. [20]. H. I. Bulbul, I. Batmaz, and M. Ozel, “Wireless
3850–3864, 2022. network security: comparison of wep (wired equivalent
[8]. A. Rahman, C. Chakraborty, A. Anwar, M. Karim, M. privacy) mechanism, wpa (wi-fi protected access) and
Islam, D. Kundu, Rahman, S. S. Band et al., “Sdn–iot rsn (robust security network) security proto-cols,” in
empowered intelligent frame-work for industry 4.0 Proceedings of the 1st international conference on
applications during covid-19 pandemic,” Cluster Forensic applications and techniques in
Computing, vol. 25, no. 4, pp. 2351–2368, 2022. telecommunications, information, and multimedia and
[9]. A. Rahman, U. Sara, D. Kundu, S. Islam, M. J. Islam, workshop, 2008, pp. 1–6.
M. Hasan, Rahman, and M. K. Nasir, “Distb- [21]. A. Halbouni, L.-Y. Ong, and M.-C. Leow, “Wireless
sdoindustry: Enhancing security in industry 4.0 security proto-cols wpa3: A systematic literature
services based on distributed blockchain through review,” IEEE Access, vol. 11, pp. 112 438–112 450,
software defined networking-iot enabled architecture,” 2023.
International Journal of Advanced Computer Science [22]. D. Kundu, M. M. Rahman, A. Rahman, D. Das, U. R.
and Applications, vol. 11, no. 9, 2020. Siddiqi, M. G. R. Alam, S. K. Dey, G. Muhammad,
[10]. M. J. Islam, A. Rahman, S. Kabir, A. Khatun, A. and Z. Ali, “Federated deep learning for monkeypox
Pritom, and Chowdhury, “Sdot-nfv: A distributed sdn disease detection on gan-augmented dataset,” IEEE
based security system with iot for smart city Access, 2024.
environments,” GUB Journal of Science and [23]. A. Rahman, M. J. Islam, F. A. Sunny, and M. K. Nasir,
Engineering, vol. 7, pp. 27–35, Jul. 2021. “Distblocksdn: A distributed secure blockchain based
[11]. A. Rahman, K. Hasan, D. Kundu, M. J. Islam, T. sdn-iot architecture with nfv implementation for smart
Debnath, S. S. Band, and N. Kumar, “On the icn-iot cities,” in 2019 2nd International Conference on
with federated learning integration of communication: Innovation in Engineering and Technology (ICIET),
Concepts, security-privacy issues, applications, and 2019, pp. 1–6.
future perspectives,” Future Generation Computer [24]. N. Dalal, N. Akhtar, A. Gupta, N. Karamchandani, G.
Systems, vol. 138, pp.61–88, 2023. S. Kasbekar, and J. Parekh, “A wireless intrusion
[12]. A. Rahman, M. Rahman, D. Kundu, M. R. Karim, S. detection system for 802.11 wpa3 networks,” in 2022
S. Band, and M. Sookhak, “Study on iot for sars-cov-2 14th International Conference on COMmunication
with healthcare:present and future perspective,” Systems NETworkS (COMSNETS), 2022, pp. 384–
Mathematical Biosciences and Engineering, vol. 18, 392.
no. 6, pp. 9697–9726, 2021. [25]. H. Debar, M. Dacier, and A. Wespi, “Towards a
[13]. H.-J. Liao, C.-H. R. Lin, Y.-C. Lin, and K.-Y. Tung, taxonomy of intrusion-detection systems,” Computer
“Intrusion detection system: A comprehensive networks, vol. 31, no. 8, pp. 805–822, 1999.
review,” Journal of Network and Computer [26]. C. P. Kohlios and T. Hayajneh, “A comprehensive
Applications, vol. 36, no. 1, pp. 16–24, 2013. attack flow model and security analysis for wi-fi and
[14]. S. Axelsson, “Research in intrusion-detection systems: wpa3,” Electronics, vol. 7, no. 11, p. 284, 2018.
A survey,” Tech-nical report 98–17. Department of [27]. N. Dalal, N. Akhtar, A. Gupta, N. Karamchandani, G.
Computer Engineering, Chalmers . . . , Tech. Rep., S. Kasbekar, and J. Parekh, “A wireless intrusion
1998. detection system for 802.11 wpa3 networks,” in 2022
[15]. M. Appel and I. S. Guenther, “Wpa 3-improvements 14th international conference on COMmunication
over wpa 2 or broken again?” Network, vol. 7, pp. 1–4, systems & NETworkS (COMSNETS). IEEE, 2022,
2020. pp. 384–392.
[16]. M. Vanhoef and E. Ronen, “Dragonblood: Analyzing [28]. F. S. Alsharbaty and Q. I. Ali, “Smart electrical
the Dragonfly handshake of WPA3 and EAP-pwd,” in substation cybersecurity model based on wpa3 and
IEEE Symposium on Security & Privacy (SP). IEEE, cooperative hybrid intrusion detection system (chids),”
2020. Smart Grids and Sustainable Energy, vol. 9, no. 1, p.
[17]. K. I. Qureshi, L. Wang, L. Sun, C. Zhu, and L. Shu, 11, 2024.
“A review on design and implementation of software- [29]. A. A. Bhutta, M. u. Nisa, and A. N. Mian,
defined wlans,” IEEE Systems Journal, vol. 14, no. 2, “Lightweight real-time wifi-based intrusion detection
pp. 2601–2614, 2020. system using lightgbm,” Wireless Networks, vol. 30,
[18]. R. Saini, D. Halder, and A. M. Baswade, “Rids: Real- no. 2, pp. 749–761, 2024.
time intrusion detection system for wpa3 enabled [30]. D. Koutras, P. Dimitrellos, P. Kotzanikolaou, and C.
enterprise networks,” in GLOBECOM 2022-2022 Douligeris, “Auto-mated wifi incident detection attack
IEEE Global Communications Conference. IEEE, tool on 802.11 networks,” in 2023 IEEE Symposium
2022, pp. 43–48. on Computers and Communications (ISCC). IEEE,
2023, pp. 464–469.

IJISRT24MAR2149 www.ijisrt.com 2940

 Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

[31]. M. Thankappan, H. Rifa`-Pous, and C. Garrigues, “A [43]. M. S. I. Khan, A. Rahman, T. Debnath, M. R. Karim,
signature-based wireless intrusion detection system M. K. Nasir,S. Band, A. Mosavi, and I. Dehzangi,
framework for multi-channel man-in-the-middle “Accurate brain tumor detection using deep
attacks against protected wi-fi networks,” IEEE convolutional neural network,” Computational and
Access, 2024. Structural Biotechnology Journal, vol. 20, pp. 4733–
[32]. R. Saifan, M. Radi, H. Al-Dabbagh, and B. Mansour, 4745, 2022.
“A lightweight log-monitoring-based mitigation tool [44]. A. Rahman, A. Montieri, D. Kundu, M. Karim, M.
against wlan attacks,” 2023. Islam, S. Umme, Nascita, A. Pescape´ et al., “On the
[33]. B. Kishiyama, J. Guerrero, and I. Alsmadi, “Security integration of blockchain and sdn: Overview,
policies automation in software defined networking,” applications, and future perspectives,” Journal of
Available at SSRN 4384690, 2023. Network and Systems Management, vol. 30, no. 4, pp.
[34]. S. L. Qaddoori and Q. I. Ali, “An efficient security 1–44,
model for industrial internet of things (iiot) system [45]. A. Rahman, M. J. Islam, S. S. Band, G. Muhammad,
based on machine learning principles,” Al-Rafidain K. Hasan, and Tiwari, “Towards a blockchain-sdn-
Engineering Journal (AREJ), vol. 28, no. 1, pp. 329– based secure architecture for cloud computing in smart
340, 2023. industrial iot,” Digital Communications and Networks,
[35]. K. Uszko, M. Kasprzyk, M. Natkaniec, and P. Chołda, vol. 9, no. 2, pp. 411–421, 2023.
“Rule-based system with machine learning support for [46]. A. I. Udoy, M. A. Rahaman, M. J. Islam, A. Rahman,
detecting anomalies in 5g wlans,” Electronics, vol. 12, Z. Ali, and Muhammad, “4sqr-code: A 4-state qr code
no. 11, p. 2355, 2023. generation model for increasing data storing capacity
[36]. D. S. M. Narayana, S. B. Nookala, S. Chopra, and U. in the digital twin framework,” Journal of Advanced
Shanmugam, “An adaptive threat defence mechanism Research, 2023.
through self defending network to prevent hijacking in [47]. A. Rahman, M. J. Islam, M. Saikat Islam Khan, S.
wifi network,” in 2023 International Conference on Kabir, A. I. Pritom, and M. Razaul Karim, “Block-
Advances in Electronics, Communication, Computing sdotcloud: Enhancing security of cloud storage through
and Intelligent Information Systems (ICAECIS). IEEE, blockchain-based sdn in iot network,” in 2020 2nd
2023, pp. 133–138. International Conference on Sustainable Technologies
[37]. K. Stella, M. Menaka, C. S. Kumar, A. P. Xavier, and for Industry 4.0 (STI), 2020, pp. 1–6.
H. Sarvesh, “Detection of hotspot spoofing by [48]. A. Rahman, M. J. Islam, M. R. Karim, D. Kundu, and
monitoring network traffic,” in 2023 International S. Kabir, “An intelligent vaccine distribution process
Conference on Sustainable Computing and Smart in covid-19 pandemic through blockchain-sdn
Systems (ICSCSS). IEEE, 2023, pp. 794–800. framework from bangladesh perspective,” in 2021
[38]. S.-E. Mansour, A. Sakhi, L. Kzaz, and A. Sekkaki, Inter-national Conference on Electronics,
“Enhancing security mechanisms for iot-fog Communications and Information Technology
networks,” Journal of Robotics and Control (JRC), vol. (ICECIT), 2021, pp. 1–4.
5, no. 1, pp. 152–159, 2024. [49]. A. Rahman, M. A. H. Wadud, M. J. Islam, D. Kundu,
[39]. S. Islam, U. Sara, A. Kawsar, A. Rahman, D. Kundu, T. A.-U.-H. Bhuiyan, G. Muhammad, and Z. Ali,
D. D. Dipta, A. R. Karim, and M. Hasan, “Sgbba: An “Internet of medical things and blockchain-enabled
efficient method for prediction system in machine patient-centric agent through sdn for remote patient
learning using imbalance dataset,” International monitoring in 5g network,” Scientific Reports, vol. 14,
Journal of Advanced Computer Science and no. 1, p. 5297, 2024.
Applications, vol. 12, no. 3, 2021. [50]. A. Rahman, J. Islam, D. Kundu, R. Karim, Z. Rahman,
[40]. A. Rahman, M. S. Hossain, G. Muhammad, D. Kundu, S. S. Band, Sookhak, P. Tiwari, and N. Kumar,
T. Debnath, Rahman, M. S. I. Khan, P. Tiwari, and S. “Impacts of blockchain in software-defined internet of
S. Band, “Federated learning-based ai approaches in things ecosystem with network function virtualization
smart healthcare: concepts, taxonomies, challenges and for smart applications: Present perspectives and future
open issues,” Cluster computing, vol. 26, no. 4, pp. directions,” International Journal of Communication
2271– 2311, 2023. Systems, p. e5429, 2023.
[41]. S. I. Khan, A. Shahrior, R. Karim, M. Hasan, and A. [51]. M. Hasan, A. Rahman, M. R. Karim, M. S. I. Khan,
Rahman, “Multinet: A deep neural network approach and M. J. Islam, “Normalized approach to find optimal
for detecting breast cancer through multi-scale feature number of topics in latent dirichlet allocation (lda),” in
fusion,” Journal of King Saud University-Computer Proceedings of International Conference on Trends in
and Information Sciences, vol. 34, no. 8, pp. 6217– Computational and Cognitive Engineering. Springer,
6228, 2022. 2021, pp. 341– 354.
[42]. T. Debnath, M. M. Reza, A. Rahman, A. Beheshti, S. [52]. K. M. Shayshab Azad, N. Hossain, M. J. Islam, A.
S. Band, and Alinejad-Rokny, “Four-layer ConvNet to Rahman, and S. Kabir, “Preventive determination and
facial emotion recognition with minimal epochs and avoidance of ddos attack with sdn over the iot
the significance of data diversity,” Scientific Reports, networks,” in 2021 International Conference on Au-
vol. 12, no. 1, p. 6991, dec 2022. [Online]. Available: tomation, Control and Mechatronics for Industry 4.0
https://www.nature.com/articles/s41598-022-11173-0 (ACMI), 2021, pp. 1–6.

IJISRT24MAR2149 www.ijisrt.com 2941

 Volume 9, Issue 3, March – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAR2149

[53]. A. Rahman, K. Hasan, and S. Jeong, “An enhanced

security architecture for industry 4.0 applications based
on software-defined networking,” in 2022 13th
International Conference on Information and
Communication Technology Convergence (ICTC),
2022, pp. 2127–2130.
[54]. A. Rahman, T. Debnath, D. Kundu, M. S. I. Khan, A.
A. Aishi, Sazzad, M. Sayduzzaman, and S. S. Band,
“Machine learning and deep learning-based approach
in smart healthcare: Recent advances, applications,
challenges and opportunities,” AIMS Public Health,
vol. 11, no. 1, pp. 58–109, 2024.
[55]. M. A. Rahaman, K. U. Oyshe, P. K. Chowdhury, T.
Debnath, A. Rahman, and M. S. I. Khan, “Computer
vision-based six layered convneural network to
recognize sign language for both numeral and alphabet
signs,” Biomimetic Intelligence and Robotics, vol. 4,
no. 1, p. 100141, 2024.
[56]. M. Tavallaee, E. Bagheri, W. Lu, and A. A. Ghorbani,
“A detailed analysis of the kdd cup 99 data set,” in
2009 IEEE symposium on computational intelligence
for security and defense applications. Ieee, 2009, pp.
1–6.
[57]. S. N. Nobel, M. A. H. Wadud, A. Rahman, D. Kundu,
A. A. Aishi, Sazzad, M. Rahman, M. A. Imran, O. F.
Sifat, M. Sayduzzaman et al., “Categorization of
dehydrated food through hybrid deep transfer learning
techniques,” Statistics, Optimization & Information
Computing, 2024.

IJISRT24MAR2149 www.ijisrt.com 2942