An Interpretable GTC-Based Framework For Balanced IoT Device Classification And Real-Time Behavioural Analytics

Read full text Summarise with VIRAT
Authors: Ayesha Banu, Embari Raghuvaran, Gankidi Praveen, Balla Ganesh, Bollu Srikanth, Amarlapudi Karunya Keerthan
Cited by: 0
References: 5

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

With the proliferation of Internet of Things (IoT) devices in smart environments, identifying and securing heterogeneous network entities has become a paramount challenge.This project presents a robust IoT Device Classification Platform that leverages supervised machine learning to categorize devices based on their network traffic fingerprints.Unlike traditional rule-based systems that struggle with the diversity of IoT hardware, this system utilizes a data-driven approach to map behavioral signatures such as packet frequency, byte counts, and communication duration to specific device profiles.A significant hurdle in IoT classification is class imbalance, where common devices overshadow rarer hardware in the dataset.To address this, the project implements Adaptive Synthetic (ADASYN) oversampling, which intelligently generates synthetic data points for minority classes, ensuring a balanced and unbiased training phase.The core of the platform is a comparative machine learning pipeline featuring Gaussian and Multinomial Naive Bayes (GNB and MNB) classifiers, a Decision Tree Classifier (DTC), and a proposed Greedy Tree Classifier (GTC) model.The GTC model, optimized via the imodels library, provides a unique balance of high-accuracy classification and human-interpretable decision rules.Developed using a modular Flask architecture, the platform offers a dual interface for Exploratory Data Analysis (EDA) and real-time inference.Experimental results demonstrate that the GTC model achieves superior F1 scores compared to baseline probabilistic models, effectively distinguishing between devices like smart cameras, sensors, and hubs.The final system provides an end-toend lifecycle from raw data ingestion to persistent model deployment, offering a scalable solution for automated network management and enhanced IoT security.

Fields of Study & Subjects

Internet Traffic Analysis and Secure E-voting Network Security and Intrusion Detection Internet of Things and AI

Publication details

Journal / source
International journal of engineering science and advanced technology.
Volume
26
Issue
3
Pages
453–463
Publication date
1 March 2026
Language
EN