An Anomaly Intrusion Detection for High-Density Internet of Things Wireless Communication Network Based Deep Learning...

Area of Science:

• Computer Science
• Cybersecurity
• Network Engineering

Background:

• Telecommunication networks are expanding rapidly, supporting diverse applications requiring secure data transmission.
• Internet-of-Things (IoT) devices, while integral to modern communication, are particularly vulnerable to cyberattacks due to their physical distribution.
• Intrusion Detection Systems (IDS) are crucial for safeguarding these networks against service disruption and data breaches.

Purpose of the Study:

• To propose and evaluate two distinct deep learning models for enhancing intrusion detection in telecommunication networks, specifically targeting IoT vulnerabilities.
• To compare the efficacy of a custom Convolutional Neural Network (CNN) combined with Long Short Term Memory (LSTM) layers against a fully connected Artificial Neural Network (ANN) model.
• To assess the performance of these models against a baseline Logistic Regression (LR) algorithm.

Main Methods:

• Development of a hybrid deep learning model integrating Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) layers.
• Construction of a second model based entirely on fully connected (dense) layers, forming a custom Artificial Neural Network (ANN).
• Comparative analysis of the proposed CNN-LSTM and ANN models against the Logistic Regression (LR) algorithm using accuracy as the primary metric.

Main Results:

• The custom Artificial Neural Network (ANN) model achieved a high accuracy of 97.01%.
• The hybrid Convolutional Neural Network-Long Short Term Memory (CNN-LSTM) model demonstrated strong performance with 96.08% accuracy.
• Both proposed deep learning models significantly outperformed the Logistic Regression (LR) algorithm, which achieved 92.8% accuracy.

Conclusions:

• The proposed Artificial Neural Network (ANN) and hybrid CNN-LSTM models offer superior intrusion detection capabilities for telecommunication networks compared to traditional methods like Logistic Regression.
• Deep learning approaches, particularly the fully connected ANN model, are highly effective in identifying and mitigating security threats in vulnerable IoT environments.
• The findings underscore the importance of advanced IDS solutions for securing the expanding landscape of connected devices and communication infrastructure.