Adaptive Machine Learning for Real‑Time Intrusion Detection in IoT

Ouku Bhulakshmi, Muddam Anusha, Ramisetty Somesh, Surasetty Badrinath,

Mangali Madhan Gopal and Pattan Thoufiq Khan

Department of Computer Science and Engineering, Santhiram Engineering College, Nandyal 518501, Andhra Pradesh,

India

Keywords: Attack Process Analysis, Internet of Things (IoT) Malware, Machine Learning (ML) Algorithms, Network

Traffic Classification, Semantic‑Level Features.

Abstract: The Internet of Things (IoT) gadgets are extensively used throughout several domains, providing numerous

conveniences to individuals' lives. However, the extensive deployment of IoT devices has made maintaining

these systems against cyber-attacks a primary concern for researchers. IoT devices possess limited computing

capabilities and storage resources, leading to inadequate security defense mechanisms and heightened

vulnerability to malware and device assaults. Current IoT-focused intrusion detection solutions often merely

identify specific malicious attempts or require intricate models and substantial processing resources to achieve

elevated detection accuracy. In this study, we utilize three datasets: BoT-IoT, MedBIoT, and MQTT-IoT-IDS

2020. We implement various algorithms, including Decision Tree, Random Forest, KNN, XGBoost, DNN,

CNN, and advanced ensemble techniques such as Stacking Classifier (DT + RF with LightGBM) and CNN

+ LSTM. Our results demonstrate that the Stacking Classifier achieved the highest performance, with superior

accuracy, precision, recall, and F1 score. The Stacking Classifier achieved a high accuracy of 100% in BoT-

IoT and MedBIoT, and 92.3% in MQTT-IoT-IDS 2020, effectively enhancing the robustness and accuracy

of IoT intrusion detection in resource-constrained environments. This method provides a lightweight and

efficient solution to improve security measures for IoT devices.

1 INTRODUCTION

The Internet of Things (IoT) is the network of

billions of devices worldwide that collect and share

data in real time. With the rapid evolution of

technology, the adoption of IoT is gaining

momentum, and more and more devices are

connecting daily. IoT has developed dramatically in

domains such as transportation, health care, industry

and smart city.

Azimjonov, J., & Kim, T. (2024). The

number of total IoT connections worldwide was 12.2

billion active endpoints in 2021, as reported by

Spring 2022. Amidst obstacles such as the COVID-

19 pandemic and a faltering supply chain, the IoT

market continued to develop with researchers

forecasting an 18% increase to 14.4 billion active

connections by 2022. Going forward, loosening

supply chain bottlenecks and an additional

acceleration of adoption may contribute to around

27bn connected IoT devices by 2025 (

Li et al., 2024).

Despite the benefits of the rapid spread of IoT

devices, it raises crucial privacy and security

concerns

Azimjonov, J., & Kim, T. (2024). Increasing

number of Internet of Things (IoT) deployments have

significantly resulted in risks that are associated with

data breaches and cybercrimes, which have

witnessed the loss of personal and corporate

information. In 2021, an Israeli cyber-security

company SAM disclosed more than 900 million IoT

attacks in that year (

Li et al., 2024). These are more

examples of the growing vulnerabilities of IoT

systems facing cyber threats.

IoT devices usually have constrained memory,

computation and communication resources, thus

complex and strong security protection is also hard

to be supported (

Fatima et al., 2024). Numerous

devices pass the un-encrypted network data amongst

devices over the wireless networks, thus attackers can

intercept, read, and analyze communications. Apart

from eavesdropping, attackers also create illicit

traffic to break IoT devices (

Tiwari et al., 2024). Such

654

Bhulakshmi, O., Anusha, M., Somesh, R., Badrinath, S., Gopal, M. M. and Khan, P. T.

Adaptive Machine Learning for Real-Time Intrusion Detection in IoT.

DOI: 10.5220/0013887900004919

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 2, pages

654-662

ISBN: 978-989-758-777-1

malicious behavior may cause device to malfunction,

provide illegal access, or even turning into a zombie

by an adversary which uses the heterogeneity of the

IoT for large scale attack.

A prime example of IoT vulnerabilities was seen

in the 2016 Distributed Denial-of-Service (DDoS)

attack launched against Dyn, a major service

provider. The attack, facilitated by the Mirai botnet,

exploited weakly secured IoT devices, leading to

significant disruptions and outages across several

services

Otokwala, U., Petrovski, A., & Kalutarage, H.

(2024). Such incidents highlight the urgent need for

robust security measures to protect IoT networks. As

IoT continues to expand, addressing these

vulnerabilities remains a critical area of focus for

researchers and developers alike (

Almotairi, A et al.,

2024)

2 RELATED WORK

In their study, (Wardana et al., 2024) proposed a

lightweight, trust-managing, and privacy-preserving

collaborative intrusion detection system (IDS) for IoT

networks. Their system emphasizes collaboration

between nodes to improve detection efficiency while

preserving user privacy. By incorporating trust

management mechanisms, the proposed model

mitigates the impact of malicious nodes, ensuring

robust intrusion detection. The system achieves

lightweight performance through computationally

efficient algorithms and is particularly suited for

resource-constrained IoT environments. This work

contributes significantly to balancing security and

efficiency in distributed IoT networks.

Ramesh Kumar and Sudhakaran focused on

enhancing IoT network security by leveraging feature

selection techniques and the Light Gradient Boosting

Machine (LGBM). Their robust intrusion detection

system optimizes the computational overhead of

feature processing while maintaining high accuracy

in detecting intrusions. The authors utilized dataset

preprocessing methods and LGBM’s capability to

handle large-scale datasets effectively. The results

showed improved performance compared to

conventional machine learning techniques, making

the system a promising solution for securing IoT

networks.

Gowthami and Vigenesh introduced a lightweight

pyramidal U-Net architecture with a tri-level dual

inception-based framework for distributed intrusion

detection in IoT networks. This innovative

framework leverages the pyramidal U-Net’s

hierarchical feature extraction capabilities combined

with dual inception modules for efficient intrusion

detection. The authors emphasized the distributed

nature of their approach, allowing scalability and

adaptability to varying IoT network sizes. Their work

stands out for its advanced architecture, which

achieves superior detection rates with reduced

resource consumption.

Francis et al. proposed a hybrid intrusion

detection approach based on the Message Queuing

Telemetry Transport (MQTT) protocol for industrial

IoT. Their model addresses the unique vulnerabilities

of the MQTT protocol, which is widely used in IoT

communication. The hybrid approach combines

anomaly-based and signature-based detection

methods, ensuring comprehensive protection against

known and emerging threats. The authors validated

their approach using real-world MQTT datasets,

achieving high accuracy and low false-positive rates.

This work is particularly relevant for IoT applications

relying on MQTT.

Vyšniūnas et al. proposed a risk-based system-call

sequence grouping method for detecting malware

intrusions in IoT networks. Their approach analyzes

system call sequences to identify anomalous behavior

indicative of malware. By grouping system calls

based on risk levels, the model enhances the precision

of intrusion detection while reducing computational

complexity. The authors demonstrated the efficacy of

their method using real-world malware datasets,

showcasing its applicability to IoT environments

where traditional signature-based methods may fail.

Musthafa et al. developed a novel intrusion

detection system optimized for IoT by combining

balanced class distribution, feature selection, and

ensemble machine learning techniques. Recognizing

the challenges posed by imbalanced datasets, the

authors applied class balancing techniques to improve

detection accuracy for minority attack types.

Additionally, their feature selection approach reduces

computational overhead, and the ensemble machine

learning model achieves state-of-the-art performance.

This study offers valuable insights into handling data

imbalance in IoT intrusion detection.

Momand et al. introduced ABCNN-IDS, an

attention-based convolutional neural network

designed specifically for intrusion detection in IoT

networks. Their architecture integrates attention

mechanisms to focus on critical features, enhancing

detection accuracy and robustness. The authors

highlighted the model’s ability to handle diverse

attack types while maintaining computational

efficiency. Experimental results demonstrated the

superiority of ABCNN-IDS over traditional CNN-

Adaptive Machine Learning for Real-Time Intrusion Detection in IoT

655

based approaches, making it a promising solution for

advanced IoT security applications.

3 MATERIALS AND METHODS

The proposed system aims to develop a lightweight

and efficient intrusion detection method tailored for

Internet of Things (IoT) devices, addressing their

security vulnerabilities and limited computational

resources. We will implement various machine

learning algorithms, including Decision Tree

(

Almotairi et al., 2024), Random Forest (Wardana, A. A

et al., 2024)

, K-Nearest Neighbors (KNN) Ramesh

Kumar, M., & Sudhakaran, P. (2024)

, XGBoost

Gowthami, D., & Vigenesh, M. (2024), Deep Neural

Network (DNN)

Francis, G. T., Souri, A., & İnanç, N.

(2024)

, and Convolutional Neural Network (CNN)

Vyšniūnas, T., Čeponis, D., Goranin, N., & Čenys, A.

(2024)

. Additionally, we will incorporate advanced

ensemble techniques such as Stacking Classifier,

which combines Decision Tree and Random Forest

with LightGBM, and a hybrid CNN + Long Short-

Term Memory (LSTM) model to enhance detection

accuracy and robustness. The system will utilize

datasets such as BoT-IoT (N. Koroniotis et al., 2019),

MedBIoT (

Guerra-Manzanares et al., 2020), and MQTT-

IoT-IDS 2020 (H. Hindy et al., 2020) for training and

evaluation. By combining multiple algorithms and

ensemble methods, the proposed system seeks to

provide a comprehensive solution for identifying and

mitigating malicious activities in IoT environments

while ensuring minimal resource consumption,

thereby improving overall security measures for IoT

devices.

This architecture in figure 1 aims to detect

intrusions in IoT networks. It starts with data

visualization and label encoding on the BoT-IoT (

Koroniotis et al., 2019)

, MedBIoT (Guerra-Manzanares et

al., 2024)

and MQTT-IoT-IDS 2020 (H. Hindy et al.,

2021)

. dataset. Feature selection is then performed

followed by data processing. The processed data is

split into training and validation sets. Various

machine learning models, including Decision Tree

(

Almotairi et al., 2024) , Random Forest (Wardana, A. A

et al., 2024)

, KNN Ramesh Kumar, M., & Sudhakaran, P.

(2024), XGBoost Gowthami, D., & Vigenesh, M. (2024),

DNN Francis, G. T., Souri, A., & İnanç, N. (2024), CNN

Vyšniūnas, T., Čeponis, D., Goranin, N., & Čenys, A.

(2024).

, and a stacking classifier, are trained and

evaluated based on metrics like accuracy, precision,

recall, and F1-score.

Figure 1: Proposed Architecture.

3.1 Dataset Collection

In this analysis, three datasets are used, such as BoT-

IoT, MedBoTIoT, and MQTT-IoT-IDS 2020,

containing network traffic data with attack labels and

traffic behaviors, specifically for intrusion detection

in IoT environments.

3.1.1 BoT-IoT Dataset

Figure 2: BoT-IoT Dataset.

The dataset in figure 2 utilized is the BoT-IoT dataset

(N. Koroniotis et al., 2019), consisting of 1,000,000

entries and 46 features, capturing diverse network

traffic details. Key features include packet statistics,

byte rates, connection states, and protocol-specific

metrics. Attack labels are categorized as "DoS" with

subcategories TCP, UDP, and HTTP, with respective

counts of 615,800, 382,715, and 1,485. Non-essential

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

656

columns, including metadata and categorical attack

labels, were dropped to streamline the dataset for

analysis. Missing values were handled by dropping

rows with null entries, ensuring data consistency.

3.1.2 MedBIoT

Figure 3: MedBoT.

The dataset in figure 3, MedBoTIoT (Guerra-

Manzanares et al., 2020), comprises network traffic data

from multiple CSV files, including *MQTT-DDoS-

Connect Flood*, *MQTT-DoS-Connect Flood*,

*MQTT-Malformed Data*, and *Benign* traffic.

Each file represents specific traffic behavior, labeled

as "DDoS," "DoS," "Malformed," and "Benign,"

respectively, by adding a target column. These

datasets were combined using the `concat()` method

to create a unified dataset for analysis. The data is

tailored for intrusion detection in IoT networks,

focusing on distinguishing between malicious and

benign traffic patterns.

3.1.3 MQTT-IoT-IDS 2020

Figure 4: MQTT-IoT-IDS 2020.

The MQTT-IoT-IDS 2020 (H. Hindy et al., 2021).

dataset contains 99,290 entries of network traffic

data, focusing on MQTT-based IoT communication

(figure 4). It includes features such as message

lengths, quality of service, and connection flags,

providing insights into communication patterns. The

dataset categorizes traffic into six labels: legitimate,

DoS, brute force, malformed, slowite, and flood,

enabling comprehensive analysis of both normal and

malicious activities. It is structured with 28 columns,

excluding unnecessary fields, and serves as a resource

for intrusion detection and anomaly detection

research in IoT networks.

3.2 Pre-Processing

We used pre-processing steps like data cleaning, label

encoding, feature selection, and data visualization to

prepare the dataset for analysis, ensuring accuracy,

reducing complexity, and enhancing model

performance for predictions.

Data Processing: Data processing involves

preparing raw data for analysis by cleaning,

transforming, and organizing it to ensure accuracy

and reliability. This includes handling missing values,

removing irrelevant features, encoding categorical

variables into numerical formats, and scaling data for

uniformity. Feature selection techniques are applied

to identify the most important variables, reducing

complexity and improving model performance. Data

processing ensures the dataset is structured,

consistent, and optimized for machine learning

models, enabling effective pattern recognition,

prediction, and decision-making during analysis.

Data Visualization: The dataset's subcategory

distribution is visualized using a bar chart with a

vibrant color palette, providing a clear comparison of

data points across different classes. This highlights

the frequency of benign and malignant nodules for

better understanding of class imbalances.

Additionally, a heatmap is used to display the

correlation between features, offering a detailed view

of relationships within the dataset. The colorful

representation enhances the interpretability of feature

interactions, aiding in identifying significant

attributes relevant to the analysis and decision-

making process.

Label Encoding: Label encoding is applied to

transform categorical values into numerical ones,

ensuring the dataset is suitable for machine learning

algorithms. By converting subcategory labels into

numeric representations, the process simplifies the

data while retaining its meaningful structure. This

transformation assigns unique integers to each

category, allowing models to process and analyze the

data more effectively. Label encoding is particularly

useful for handling non-numeric columns, making

Adaptive Machine Learning for Real-Time Intrusion Detection in IoT

657

them compatible with computational methods while

preserving their inherent distinctions for accurate

analysis and prediction.

Feature Selection: Feature selection is performed to

identify and retain the most relevant variables that

significantly influence the target outcome, improving

model performance and reducing complexity. By

applying a percentile-based selection method, the

process evaluates the mutual information between

features and the target variable, ensuring only the top

25% of impactful features are retained. This approach

enhances the dataset by focusing on critical

predictors, eliminating irrelevant or redundant

variables, and enabling the model to achieve better

accuracy, efficiency, and interpretability in its

analysis and predictions.

3.3 Training and Testing

The dataset is divided into two parts: one for training

and another for testing, ensuring an effective

evaluation of the model's performance. The training

set, which comprises 80% of the data, is used to teach

the model by identifying patterns and relationships

between input features and the target variable. The

testing set, containing 20% of the data, evaluates how

well the trained model can generalize its predictions

on unseen data. This balanced split ensures accuracy

and reliability in assessing model performance.

3.4 Algorithms

A Decision Tree classifies incoming data as benign

or malicious by creating a series of rules based on

feature values. Its interpretability makes it effective

for identifying threats and understanding decision-

making processes in IoT environments (

Almotairi et

al., 2024).

Random Forest constructs multiple decision trees

and aggregates their predictions for classification. It

enhances accuracy, reduces overfitting, and handles

diverse IoT data, detecting complex attack patterns

with high performance and generalization capabilities

(

Wardana, A. A et al., 2024).

K-Nearest Neighbors (KNN) classifies data points

based on the majority class of their nearest neighbors.

It detects potential threats by comparing features of

incoming data with labeled examples, adapting well

to changes in data distribution in real-time

environments

Ramesh Kumar, M., & Sudhakaran, P.

(2024).

XGBoost is a gradient boosting algorithm that

improves speed and performance. It handles large

datasets and complex feature interactions,

minimizing loss functions and preventing overfitting,

ensuring robust and adaptive detection of evolving

cyber threats in IoT systems

Gowthami, D., & Vigenesh,

M. (2024).

Deep Neural Network (DNN) processes high-

dimensional data to learn complex patterns. It

enhances detection of sophisticated attacks,

improving accuracy in identifying malicious

activities by capturing intricate relationships between

features and handling diverse data types effectively

Francis, G. T., Souri, A., & İnanç, N. (2024).

Convolutional Neural Network (CNNs) analyze

structured grid data, such as images or signals. They

automatically extract relevant features, improving

anomaly detection and real-time threat identification,

making them highly effective for detecting attacks in

IoT environments through pattern recognition

Vyšniūnas, T., Čeponis, D., Goranin, N., & Čenys, A.

(2024).

Stacking Classifier (DT + RF with LightGBM):

The Stacking Classifier integrates Decision Trees,

Random Forest, and LightGBM to improve

prediction accuracy. By combining strengths from

each model, it reduces false positives and negatives,

creating a robust system for threat identification in

IoT devices.

4 RESULTS & DISCUSSION

Accuracy: The true accuracy of a test is the

proportion of the test to be able to correctly identify

patient and healthy subjects. When we want to

estimate the accuracy of a test, we need to compute

the ratio of true positive and true negative cases over

all tested cases. This can be expressed

mathematically as:

Accuracy 

TPTN

TPFPTNFN

(1)

Precision: Precision measures the proportion of true

positives among the samples identified as positive.

Therefore, the precision formula is expressed as:

Precision 

True Positive

True Positive False Positive

(2)

Recall: Recall is a machine learning metric that

calculates the model’s capability to find all relevant

objects of a class. It represents how successful the

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

658

classification was in predicting the positive class, in

relation to the actual positive instances.

Recall 

TP  FN

(3)

F1-Score: F1 score is an ML evaluation metric which

gives an idea about how good your model is. It is a

combination of precision and recall for a model.

Accuracy measures how many times the model made

correct predictions over all predictions made.

𝐹1 𝑆𝑐𝑜𝑟𝑒  2 ∗

 X 



∗ 100 (4)

In Table (1, 2 and 3) the Stacking Classifier

consistently achieved the highest accuracy,

outperforming all models across all datasets and

metrics.

Table 1: Performance Evaluation Table – BoT-IoT.

ML Model Accuracy Precision Recall F1_score

KNN 1.000 1.000 1.000 1.000

DecisionTree 1.000 1.000 1.000 1.000

RandomForest 1.000 1.000 1.000 1.000

XGBoost 1.000 1.000 1.000 1.000

Extension 1.000 1.000 1.000 1.000

DNN 0.622 0.073 0.270 0.115

CNN 0.993 0.995 0.993 0.994

CNN+LSTM 0.616 1.000 0.616 0.762

Table 2: Performance Evaluation Table – MedBIoT.

ML Model Accuracy Precision Recall F1_score

KNN 0.991 0.994 0.991 0.992

DecisionTree 0.997 0.997 0.997 0.997

RandomForest 0.998 0.998 0.998 0.998

XGBoost 0.998 0.998 0.998 0.998

Extension 1.000 1.000 1.000 1.000

DNN 0.504 0.198 0.445 0.274

CNN 0.975 0.992 0.975 0.983

CNN+LSTM 0.923 0.979 0.923 0.949

Table 3: Performance Evaluation Table – UNSW-NB15 – With SMOTEENN.

ML Model Accuracy Precision Recall F1_score

KNN 0.901 0.914 0.901 0.906

Decision Tree 0.907 0.922 0.907 0.912

Random Forest 0.907 0.922 0.907 0.913

XGBoost 0.908 0.925 0.908 0.914

Extension 0.923 0.934 0.923 0.926

DNN 0.393 0.002 0.043 0.004

CNN 0.793 0.895 0.793 0.821

CNN+LSTM 0.797 0.909 0.797 0.826

Adaptive Machine Learning for Real-Time Intrusion Detection in IoT

659

Figure 5: Comparison Graph – NSL-KDD – Without SMOTEENN.

Figure 6: Comparison Graph – BoT-IoT.

Figure 7: Comparison Graph – MQTT-IoT-IDS 2020.

0.5

1.5

Chart Title

Accuracy Precision Recall F1_score

0.2

0.4

0.6

0.8

1.2

Chart Title

Accuracy Precision Recall F1_score

0.2

0.4

0.6

0.8

Comparison Graph

Accuracy Precision Recall F1_score

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

660

In figure (5, 6 & 7) accuracy is represented in light

blue, precision in orange, recall in grey and F1 Score

in yellow. The Graphs illustrate the Stacking

Classifier's superior performance across all metrics

and datasets, consistently achieving the highest

accuracy, demonstrating its robustness and

effectiveness in intrusion detection.

5 CONCLUSIONS

In conclusion, this study highlights the critical

importance of robust and efficient intrusion detection

systems (IDS) for securing Internet of Things (IoT)

devices, which are increasingly vulnerable due to

their limited computational and storage resources.

The research explores a variety of machine learning

algorithms, including Decision Tree (

Almotairi et al.,

2024), Random Forest (Wardana, A. A et al., 2024),

KNN Ramesh Kumar, M., & Sudhakaran, P. (2024).,

XGBoost

Gowthami, D., & Vigenesh, M. (2024), DNN

Francis, G. T., Souri, A., & İnanç, N. (2024), CNN

Vyšniūnas, T., Čeponis, D., Goranin, N., & Čenys, A.

(2024).

, and advanced ensemble methods like

Stacking Classifier (DT + RF with LightGBM) and

CNN + LSTM, using datasets from BoT-IoT (

Koroniotis et al., 2019), MedBIoT (Guerra-Manzanares et

al., 2020), and MQTT-IoT-IDS 2020 (H. Hindy et al.,

2021)

. The results reveal that the Stacking Classifier,

combining the strengths of multiple models,

outperforms individual classifiers, achieving

remarkable detection performance. It achieved 100%

accuracy on the BoT-IoT and MedBIoT datasets, and

92.3% accuracy on MQTT-IoT-IDS 2020. These

findings demonstrate that the Stacking Classifier

provides a highly effective, lightweight, and efficient

solution for IoT intrusion detection, significantly

enhancing security in resource-constrained

environments. This method addresses the challenges

posed by IoT devices' limitations while ensuring high

detection accuracy, thereby making a substantial

contribution to improving IoT security in practical

applications.

The future scope of this study includes exploring

more advanced ensemble techniques and hybrid

models to further enhance detection accuracy and

reduce computational overhead. Additionally, the

integration of real-time detection systems,

incorporating adaptive learning algorithms, can

improve the responsiveness to emerging threats.

Future research could also focus on incorporating

anomaly detection and federated learning to address

privacy concerns, enabling scalable and robust

security solutions for diverse IoT environments.

REFERENCES

A. Guerra-Manzanares, J. Medina-Galindo, H. Bahsi, and

S. Nõmm, “MedBIoT: generation of an IoT botnet

dataset in a medium-sized IoT network,” in Proc. 6th

Int. Conf. Inf. Syst. Security Privacy, 2020, pp. 207–

218.

Almotairi, A., Atawneh, S., Khashan, O. A., & Khafajah,

N. M. (2024). Enhancing intrusion detection in IoT

networks using machine learning-based feature

selection and ensemble models. Systems Science &

Control Engineering, 12(1), 2321381.

Azimjonov, J., & Kim, T. (2024). Designing accurate

lightweight intrusion detection systems for IoT

networks using fine-tuned linear SVM and feature

selectors. Computers & Security, 137, 103598.

Azimjonov, J., & Kim, T. (2024). Stochastic gradient

descent classifier-based lightweight intrusion detection

systems using the efficient feature subsets of datasets.

Expert Systems with Applications, 237, 121493.

Chaitanya, V. Lakshmi, and G. Vijaya Bhaskar. "Apriori vs

Genetic algorithms for Identifying Frequent Item Sets."

International journal of Innovative Research &

Development 3.6 (2014): 249-254.

Chaitanya, V. Lakshmi. "Machine Learning Based

Predictive Model for Data Fusion Based Intruder Alert

System." journal of algebraic statistics 13.2 (2022):

2477-2483.

Chaitanya, V. Lakshmi, et al. "Identification of traffic sign

boards and voice assistance system for driving." AIP

Conference Proceedings. Vol. 3028. No. 1. AIP

Publishing, 2024.

Devi, M. Sharmila, et al. "Machine Learning Based

Classification and Clustering Analysis of Efficiency of

Exercise against Covid-19 Infection." Journal of

Algebraic Statistics 13.3 (2022): 112-117.

Devi, M. Sharmila, et al. "Extracting and Analyzing

Features in Natural Language Processing for Deep

Learning with English Language." Journal of Research

Publication and Reviews 4.4 (2023): 497-502.

Fatima, M., Rehman, O., Ali, S., & Niazi, M. F. (2024).

ELIDS: Ensemble Feature Selection for Lightweight

IDS against DDoS Attacks in Resource-Constrained

IoT Environment. Future Generation Computer

Systems, 159, 172-187.

Francis, G. T., Souri, A., & İnanç, N. (2024). A hybrid

intrusion detection approach based on message queuing

telemetry transport (MQTT) protocol in industrial

internet of things. Transactions on Emerging

Telecommunications Technologies, 35(9), e5030.

Gowthami, D., & Vigenesh, M. (2024). Distributed and

Lightweight Intrusion Detection for IoT: A Lightweight

Pyramidal U-Net with Tri-Level Dual Inception-Based

Framework. In the Convergence of Self-Sustaining

Systems with AI and IoT (pp. 154-173). IGI Global.

H. Hindy, E. Bayne, M. Bures, R. Atkinson, C. Tachtatzis,

and X. Bellekens, “Machine learning based IoT

intrusion detection system: An MQTT case study

(MQTT-IoT-IDS2020 dataset),” in Proc. 12th Int.

Netw. Conf. Sel. Papers, 2021, pp. 73–84.

Adaptive Machine Learning for Real-Time Intrusion Detection in IoT

661

Li, J., Othman, M. S., Chen, H., & Yusuf, L. M. (2024).

Optimizing IoT intrusion detection system: feature

selection versus feature extraction in machine learning.

Journal of Big Data, 11(1), 36.

Li, J., Chen, H., Shahizan, M. O., & Yusuf, L. M. (2024).

Enhancing IoT Security: A Comparative Study of

Feature Reduction Techniques for Intrusion Detection

System. Intelligent Systems with Applications, 200407.

Mahammad, Farooq Sunar, Karthik Balasubramanian, and

T. Sudhakar Babu. "A comprehensive research on

video imaging techniques." All Open Access, Bronze

(2019).

Mahammad, Farooq Sunar, and V. Madhu Viswanatham.

"Performance analysis of data compression algorithms

for heterogeneous architecture through parallel

approach." The Journal of Supercomputing 76.4

(2020): 2275-2288.

Mahammad, Farooq Sunar, et al. "Key distribution scheme

for preventing key reinstallation attack in wireless

networks." AIP Conference Proceedings. Vol. 3028.

No. 1. AIP Publishing, 2024.

Mandalapu, Sharmila Devi, et al. "Rainfall prediction using

machine learning." AIP Conference Proceedings. Vol.

3028. No. 1. AIP Publishing, 2024.

Momand, A., Jan, S. U., & Ramzan, N. (2024). ABCNN-

IDS: attention-based convolutional neural network for

intrusion detection in IoT networks. Wireless Personal

Communications, 136(4), 1981-2003.

Mr.M.Amareswara Kumar, "Baby care warning system

based on IoT and GSM to prevent leaving a child in a

parked car" in International Conference on Emerging

Trends in Electronics and Communication Engineering

- 2023, API Proceedings July-2024.

Mr.M.Amareswara Kumar, Effective Feature Engineering

Technique For Heart Disease Prediction With Machine

Learning" in International Journal of Engineering &

Science Research, Volume 14, Issue 2, April-2024 with

ISSN 2277-2685.

Musthafa, M. B., Huda, S., Kodera, Y., Ali, M. A., Araki,

S., Mwaura, J., & Nogami, Y. (2024). Optimizing iot

intrusion detection using balanced class distribution,

feature selection, and ensemble machine learning

techniques. Sensors, 24(13), 4293.

N. Koroniotis, N. Moustafa, E. Sitnikova, and B. Turnbull,

“Towards the development of realistic botnet dataset in

the Internet of Things for network forensic analytics:

Bot-IoT dataset,” Future Gener. Comput. Syst., vol.

100, pp. 779–796, Nov. 2019. [Online]. Available:

https:// www.sciencedirect.com/science/article/pii/S0

167739X18327687

Otokwala, U., Petrovski, A., & Kalutarage, H. (2024).

Optimized common features selection and deep-

autoencoder (OCFSDA) for lightweight intrusion

detection in Internet of things. International Journal of

Information Security, 1-23.

Paradesi Subba Rao, "Detecting malicious Twitter bots

using machine learning" AIP Conf. Proc. 3028, 020073

(2024), https://doi.org/10.1063/5.0212693.

Paradesi Subba Rao, "Morphed Image Detection using

Structural Similarity Index Measure"M6 Volume 48

Issue 4 (December 2024), https://powertechjournal.C

Parumanchala Bhaskar, et al. "Incorporating Deep Learning

Techniques to Estimate the Damage of Cars During the

Accidents" AIP Conference Proceedings. Vol. 3028.

No. 1. AIP Publishing, 2024.

Parumanchala Bhaskar, et al "Cloud Computing Network

in Remote Sensing-Based Climate Detection.

Parumanchala Bhaskar, et al. "Machine Learning Based

Predictive Model for Closed Loop Air.

Ramesh Kumar, M., & Sudhakaran, P. (2024). Securing IoT

networks: A robust intrusion detection system

leveraging feature selection and LGBM. Peer-to-Peer

Networking and Applications, 1-23.

Suman, Jami Venkata, et al. "Leveraging natural language

processing in conversational Al agents to improve

healthcare security." Conversational Artificial

Intelligence (2024): 699-711.

Sunar, Mahammad Farooq, and V. Madhu Viswanatham.

"A fast approach to encrypt and decrypt of video

streams for secure channel transmission." World

Review of Science, Technology and Sustainable

Development 14.1 (2018): 11-28.

Tiwari, R. S., Lakshmi, D., Das, T. K., Tripathy, A. K., &

Li, K. C. (2024). A lightweight optimized intrusion

detection system using machine learning for edge-

based IIoT security. Telecommunication Systems, 1-

20.

Vyšniūnas, T., Čeponis, D., Goranin, N., & Čenys, A.

(2024). Risk-Based System-Call Sequence Grouping

Method for Malware Intrusion Detection. Electronics,

13(1), 206.

Wardana, A. A., Kołaczek, G., & Sukarno, P. (2024).

Lightweight, Trust-Managing, and Privacy-Preserving

Collaborative Intrusion Detection for Internet of

Things. Applied Sciences, 14(10), 4109.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

662