PAPER ID: ICOEI-070

EMPOWERING IOT IN CYBER NETWORK

ATTACKS USING MACHINE LEARNING
ALGORITHMS.
By:
Harini M(312421205025)
Kaviya K(312421205043)
Dr. Niroshini Infantia H(Associate Professor)

DATE : 24TH April.

 OBJECTIVE

• The project investigates the dual role of machine learning (ML) in Internet of Things
(IoT) cybersecurity.
• It aims to enhance IoT security by analyzing vast amounts of data from IoT devices
using ML algorithms.
• The goal is to develop ML models that detect hidden patterns in network traffic,
identify anomalies, and predict future threats.
• The project seeks to fortify IoT network defenses, ensuring a proactive approach to
minimizing potential cyber threats.
• It also aims to understand how malicious actors could use ML techniques to launch
sophisticated cyberattacks, exploiting vulerabilities in IoT devices or networks.
• The ultimate objective is to balance leveraging ML for enhanced cybersecurity with
mitigating the risks of its misuse, contributing to the development of robust IoT
cybersecurity frameworks.
 ABSTRACT

• This project investigates the dual role of machine learning as a tool for bolstering and
jeopardizing IoT cybersecurity.
• Machine learning systems can process immense volumes of IoT data, identifying
patterns in traffic, flagging anomalies indicative of cyber threats, and forecasting
vulnerabilities.
• These capabilities empower IoT networks to proactively defend against and mitigate the
impact of potential attacks.
• Attackers might deploy machine learning to locate and exploit weaknesses in IoT
systems, potentially circumventing traditional defenses.
• Leveraging machine learning for IoT security, it is crucial to address its potential for
misuse.
 LITERATURE SURVEY

S.NO Title, Author, Year Technique used Description Remarks

01 Detecting Cyber Threats in IoT Networks: A gradient boosting, It demonstrates effectiveness offers machine learning-based
Machine Learning Approach, Atheer Alaa convolutional neural networks, in enhancing cybersecurity for framework for effectively
Hammad, May Adnan Falih, Senan Ali Abd and long short-term memory IoT networks. detecting cyber threats in IoT
Saadaldeen Rashid Ahmed, 2024. networks, and recurrent neural networks.
networks .

02 A Situation Based Predictive Approach for Decision Tree, MLP, and It presents a predictive It presents a well predictive
Cybersecurity Intrusion Detection and Prevention Autoencoder models for framework using machine framework.
Using Machine Learning and Deep Learning multidimensional and binary learning and deep learning
Algorithms in Wireless Sensor Networks of classification. algorithms to enhance
Industry 4.0, Fatima Al-Quayed , Zulfiqar Ahmad, cybersecurity in Industry 4.0
and Mamoona Humayun, 2024 . wireless sensor networks.
03 Empowering Digital Resilience: Neural Networks for classification to A machine learning-based IDS Promising cybersecurity
Machine Learning-Based Policing develop an Intrusion Detection System designed to quickly detect Wi-Fi enhancement tool.
Models for Cyber-Attack Detection in (IDS). cyber-attacks for the Indonesian
Wi-Fi Network, Suryadi MT, Achmad National Police.
Eriza Aminanto and Muhamad Erza
Aminanto, 2024.

04 Machine learning in cybersecurity: A Supervised Learning, ML techniques for cybersecurity, A comprehensive review of
review of threat detection and Unsupervised Learning, focusing on their roles in threat machine learning applications
defense mechanisms, Ugochukwu Deep Learning, detection, feature engineering, in cybersecurity threat
Ikechukwu Okoli, Ogugua Chimezie Reinforcement Learning. and data pretreatment. detection and defense
Obi, Adebunmi Okechukwu Adewusi strategies.
and Temitayo Oluwaseun Abrahams,
2024.

05 Study on Empowering Cyber Security Random Forest, K-Nearest Neighbor, A robust ML-based framework explores adaptive machine
by Using Adaptive Machine Learning and Support Vector Machine achieves 96.16% accuracy in learning methods to
Methods, Hari Gonaygunta, Geeta detecting malware and phishing, strengthen cybersecurity
Sandeep Nadella, Priyanka Pramod surpassing traditional methods. through dynamic threat
Pawar, and Deepak Kumar, 2024. response.
06 Empowering Intrusion Detection In Iot Machine Learning, Deep Learning, The study utilizes ML, DL, offers effective machine
Networks With Efficient Machine Neural Networks, Explainable AI, Explainable AI, and Federated learning strategies to
Learning Strategies, Souradip Roy, Federated Learning. Learning to enhance intrusion enhance intrusion detection
2023. detection in IoT networks. in IoT networks.

07 Automated Machine Learning Enabled Mayfly Optimization (MFO) with the MFO-RELM model, effectively leverages
Cybersecurity Threat Detection in Regularized Extreme Learning integrating Mayfly Automated Machine Learning
Internet of Things Environment, Fadwa Machine (RELM) Optimization with Regularized (AutoML) to enhance
Alrowais, Sami Althahabi, Saud S. Extreme Learning Machine for cybersecurity threat
Alotaibi, Abdullah Mohamed, Manar improved IoT threat detection. detection in IoT
Ahmed Hamza, and Radwa Marzouk, environments.
2023

08 Cyber Restoration of Power Systems: Mixed Integer Linear Programming uses MILP to quickly restore offers a robust methodology
Concept and Methodology for Resilient (MILP) system observability after for restoring power system
Observability, Shamsun Nahar Edib, cyber disruptions, and shows observability post-cyberattack
Yuzhang Lin, Vinod M. Vokkarane, Feng superior performance using optimization
Qiu, Rui Yao and Bo Chen, 2023. compared to heuristic techniques.
methods.
09 A data-driven approach for Synthetic Minority Oversampling It approaches using SMOTE, offers a robust AutoML-based
intrusion and anomaly detection Technique (SMOTE) for dataset mutual information, and solution for efficient intrusion
using automated machine learning quality improvement, mutual automated machine learning to and anomaly detection in IoT
for the Internet of Things, Hao Xu, information for feature selection. enhance intrusion and anomaly environments.
Zihan Sun, Yuan Cao, Hazrat Bilal, detection in IoT.
2023.

10 AI and Machine Learning in predictive analysis, behavioral highlights how AI and ML are utilizes AI and ML to
Cybersecurity: Leveraging analytics, and threat intelligence. crucial for enhancing cybersecurity strengthen cybersecurity
Technology to Combat Threats, by enabling real-time anomaly defenses against evolving
Lee Kasowaki and Koraye Emir, detection, predictive analysis, and threats.
2023. adaptive defenses against evolving
threats.
 EXISTING SYSTEM
• Cyber search engines like Shodan and Censys excel in indexing IoT devices by scanning
and fingerprinting them to map IP addresses to devices.
• In the vast address space and the dynamic nature of these mappings, managing scan
efficiency is crucial.
• Reinforcement learning offers a promising approach to optimize scanning strategies by
focusing on networks with high churn rates in IP-device mappings, though this approach
has not been extensively studied before.
• It explores this optimization problem by applying reinforcement learning to a global IoT
scanning platform, aiming to enhance scanning efficiency through a structured approach
based on real-world IoT scan data.
• Key factors influencing scanning effectiveness are identified, and practical experiments
show that the proposed system can achieve up to 40 times more IP-device mapping
updates compared to traditional random or sequential scanning methods.
 DISADVANTAGES

•High dependency on historical data for learning patterns.

•Limited visibility in private or secured networks.
•Risk of adversarial manipulation affecting scanning efficiency.
•Potential legal and ethical issues related to unauthorized scanning.
•Scalability challenges with the growing number of IoT devices.
•Inability to ensure real-time detection across all network segments.
•Occurrence of false positives and false negatives in detection.
•Computational resource intensity for large-scale deployments.
 PROPOSED SYSTEM

• In the proposed methodology a multi-layered machine learning framework for proactive

defense of IoT networks against cyberattacks is done.
• The system incorporates an anomaly detection engine utilizing recurrent neural networks
to identify aberrant network traffic patterns indicative of potential threats.
• Additionally, a threat prediction module employing support vector machines analyzes
real-time and historical data to forecast emerging attack vectors.
• Integrating these components and automating response mechanisms, the framework aims
to significantly enhance the resilience of IoT infrastructures.
• In this proposed methodology, we use Bagging Classifier, Gaussian Naïve bayes, Random
Forest algorithms.
 ADVANTAGES

•Proactive detection and classification of intrusion attempts.

•Utilizes multiple machine learning models for improved accuracy.
•Real-time monitoring and analysis of IoT network data.
•Enhanced resilience against cyber threats through adaptive learning.
•Ability to distinguish between normal and tampered data.
•Scalable framework suitable for diverse IoT environments.
•Improved threat visibility using data-driven intelligence.
 SYSTEM REQUIREMENT

Hardware Requirement:
• Processor :
Intel® Core™ i7 processor 14650HX (30M Cache, up to 5.20 GHz)
• Hard disk : minimum 80 GB
• RAM : minimum 4 GB

Software Requirement:
• Operating System : Windows
• Tool : Anaconda with Jupyter Notebook
ARCHITECTURE FOR ANOMALY DETECTION:
 MODULE DESCRIPTION:

Data Preprocessing
• Validation techniques in machine learning are used to get the error rate of the Machine
Learning (ML) model, which can be considered as close to the true error rate of the
dataset.
• If the data volume is large enough to be representative of the population, you may not
need the validation techniques.
• However, in real-world scenarios, to work with samples of data that may not be a true
representative of the population of given dataset.
• To finding the missing value, duplicate value and description of data type whether it is
float variable or integer. The sample of data used to provide an unbiased evaluation of
a model fit on the training dataset while tuning model hyper parameters.
 MODULE DESCRIPTION:

Data Analysis of Visualization

• Machine learning algorithms analyze large-scale IoT data to detect patterns, anomalies,
and potential cyber threats, enhancing security measures.
• Visualization of network traffic patterns can highlight irregularities, helping in early
detection of cyberattacks and improving threat prediction.
• Adversarial machine learning can be visualized through models showing how attackers
exploit IoT vulnerabilities to bypass security defenses.
• A comparative analysis of machine learning-based security versus traditional security
approaches can be represented using performance metrics.
• Predictive analytics visualization can showcase future cyber threats, helping in proactive
decision-making for IoT security enhancements.
 MODULE DESCRIPTION:

Deployment:
• The trained Random Forest Regressor model is saved as a pickle file to preserve its state
for future use.
• A Django project is created, with the necessary settings and configurations to handle
machine learning tasks.
• The model is loaded into the Django application using Python’s pickle module or joblib
when a prediction is required.
• An API endpoint or view is created to accept user inputs, process the data, and pass it to
the model for prediction.
• The model's prediction is returned to the user through a response integrated into a web
page.
 MODEL EVALUATION

• Gaussian Naïve Bayes: Evaluated using accuracy, precision, recall, and F1-score, it
performs well on simple, independent features but may struggle with correlated IoT
security data.
• Bagging Classifier: Assessed through cross-validation and out-of-bag (OOB) error, it
improves stability and reduces variance, enhancing anomaly detection in IoT networks.
• Random Forest: Measured by feature importance scores, confusion matrix, and AUC-
ROC curve, it provides high accuracy and robustness against overfitting in IoT
cybersecurity.
• Comparison Metrics: Precision-recall tradeoff and computational efficiency are analyzed
to determine which model balances accuracy and real-time threat detection.
• Performance Trade-offs: Gaussian Naïve Bayes excels in speed, Bagging enhances
generalization, and Random Forest offers strong feature discrimination, each with distinct
strengths in IoT security.
COMPARISON GRAPH FOR THE MODELS:

• Gaussian Naïve Bayes has significantly

lower accuracy (44.85%).Bagging
Classifier (93.86%) and Random Forest
(93.87%) perform much better, with
minimal difference between them.
INPUT PAGE
OUTPUT PAGE
 CONCLUSION
• To conclude, empowering IoT cyber networks against cyberattacks through machine
learning presents a transformative method for improving security.
• By employing cutting-edge machine learning algorithms, we can create systems capable
of detecting and countering a wide range of cyber threats, adapting to new attack
techniques with high accuracy.
• Models like anomaly detection systems and supervised classifiers provide continuous
oversight and evaluation of network behavior, identifying possible threats before they can
inflict serious damage.
• This proactive approach not only enhances the stability of IoT networks but also
guarantees that security strategies remain fluid and responsive.
• In the final analysis, incorporating machine learning into the defenses of IoT cyber
networks significantly strengthens the security framework, protecting essential
infrastructure and sensitive data from the ever-more complex and sophisticated cyber
threat landscape.
 FUTURE ENHANCEMENT

•Integration of deep learning models for advanced threat detection.

•Real-time automated mitigation of detected intrusions.
•Expansion of dataset sources for more robust model training.
•Implementation of explainable AI for better transparency.
•Incorporation of federated learning for decentralized security.
•Enhancement of adversarial attack detection and defense mechanisms.
•Development of a user-friendly dashboard for live monitoring and reporting.
 REFERENCES
• Detecting Cyber Threats in IoT Networks: A Machine Learning Approach, Atheer Alaa Hammad, May
Adnan Falih, Senan Ali Abd and Saadaldeen Rashid Ahmed, 2024.
• A Situation Based Predictive Approach for Cybersecurity Intrusion Detection and Prevention Using
Machine Learning and Deep Learning Algorithms in Wireless Sensor Networks of Industry 4.0, Fatima
Al-Quayed , Zulfiqar Ahmad, and Mamoona Humayun, 2024 .
• Empowering Digital Resilience: Machine Learning-Based Policing Models for Cyber-Attack Detection in
Wi-Fi Network, Suryadi MT, Achmad Eriza Aminanto and Muhamad Erza Aminanto, 2024.
• Machine learning in cybersecurity: A review of threat detection and defense mechanisms, Ugochukwu
Ikechukwu Okoli, Ogugua Chimezie Obi, Adebunmi Okechukwu Adewusi and Temitayo Oluwaseun
Abrahams, 2024.
• Study on Empowering Cyber Security by Using Adaptive Machine Learning Methods, Hari Gonaygunta,
Geeta Sandeep Nadella, Priyanka Pramod Pawar, and Deepak Kumar, 2024.
• Empowering Intrusion Detection In Iot Networks With Efficient Machine Learning Strategies, Souradip
Roy, 2023.
• Automated Machine Learning Enabled Cybersecurity Threat Detection in Internet of Things
Environment, Fadwa Alrowais, Sami Althahabi, Saud S. Alotaibi, Abdullah Mohamed, Manar Ahmed
Hamza, and Radwa Marzouk, 2023
• A data-driven approach for intrusion and anomaly detection using automated machine learning for the
Internet of Things, Hao Xu, Zihan Sun, Yuan Cao, Hazrat Bilal, 2023.
• AI and Machine Learning in Cybersecurity: Leveraging Technology to Combat Threats, Lee Kasowaki
and Koraye Emir, 2023.
• Cyber Restoration of Power Systems: Concept and Methodology for Resilient Observability, Shamsun
Nahar Edib, Yuzhang Lin, Vinod M. Vokkarane, Feng Qiu, Rui Yao and Bo Chen, 2023.
• Ande, R., Adebisi, B., Hammoudeh, M., & Saleem, J. (2020). Internet of Things: Evolution and
technologies from a security perspective. Sustainable Cities and Society, 54, 101728.
• Worlu, C., Jamal, A. A., & Mahiddin, N. A. (2019). Wireless sensor networks, internet of things, and
their challenges. International Journal of Innovative Technology and Exploring Engineering, 8(12S2),
556-566.
• Jurcut, A. D., Ranaweera, P., & Xu, L. (2020). Introduction to IoT security. IoT security: advances in
authentication, 27-64.
• Liang, W., & Ji, N. (2022). Privacy challenges of IoT-based blockchain: a systematic review. Cluster
Computing, 25(3), 2203- 2221.
• Jiang, X., Lora, M., & Chattopadhyay, S. (2020). An experimental analysis of security vulnerabilities in
industrial IoT devices. ACM Transactions on Internet Technology (TOIT), 20(2), 1-24.
• Djenna, A., Harous, S., & Saidouni, D. E. (2021). Internet of things meet internet of threats: New
concern cyber security issues of critical cyber infrastructure. Applied Sciences, 11(10), 4580