Deep hybrid architecture for multi-class detection of network layer attacks in WSN

Area of Science:

• Cybersecurity
• Network Security
• Artificial Intelligence

Background:

• Wireless Sensor Networks (WSNs) face significant cybersecurity threats at the network layer, impacting data integrity and network lifespan.
• Common network layer attacks include Blackhole, Flooding, and Selective Forwarding, degrading WSN performance and reliability.
• Existing detection methods struggle with the complexity and evolving nature of WSN attacks.

Purpose of the Study:

• To develop and evaluate a novel hybrid deep learning (DL) model for detecting multi-class attacks in Wireless Sensor Networks (WSNs).
• To combine the strengths of Gated Recurrent Units (GRU) and Long Short-Term Memory (LSTM) for enhanced temporal and sequential pattern recognition in WSN attack detection.
• To assess the proposed model's performance against traditional DL models using key metrics like accuracy, precision, recall, and F1-score.

Main Methods:

• Development of a hybrid deep learning model integrating Gated Recurrent Units (GRU) and Long Short-Term Memory (LSTM) architectures.
• Utilizing the WSNBFSF dataset for training and evaluating the model's capability in identifying various WSN attack types.
• Performance evaluation using metrics such as accuracy, precision, recall, and F1-score, with and without the Synthetic Minority Oversampling Technique (SMOTE) and k-fold validation.

Main Results:

• The proposed GRU+LSTM hybrid model demonstrated superior performance in multi-class WSN attack classification compared to individual DL models (DNN, RNN, ANN, Attention, LSTM, GRU).
• Achieved an overall accuracy of 97.41% without SMOTE and an improved accuracy of 98.74% with SMOTE and k-fold validation.
• Attained macro F1 and weighted F1 scores of 0.9874, indicating high effectiveness across multiple attack classes.

Conclusions:

• The GRU+LSTM hybrid model effectively captures complex sequential patterns characteristic of WSN attacks.
• The proposed architecture offers a reliable and accurate solution for intelligent intrusion detection in Wireless Sensor Network environments.
• This hybrid DL approach significantly enhances WSN security by providing robust detection capabilities against diverse network layer threats.