Hybrid Recurrent with Spiking Neural Network Model for Enhanced Anomaly Prediction in IoT Networks Security

Yasir Eltigani Ali Mustafa^1*Mohammed Mustafa Ali Mustafa²Sarah Mustafa Eljack Babiker³

• ¹Ahmed Bin Mohammed Military College, Al Rayyan, Qatar
• ²University of Tabuk, Tabuk, Saudi Arabia
• ³Majmaah University, Al Majmaah, Saudi Arabia

As the number of Internet of Things (IoT) devices grows quickly, cyber threats are becoming more complex; thus, we need better network security solutions. This research presents a combination of Neural Network (NN) models that combine Recurrent-NN (RNN) and Spiking-NN (SNN), referred to as HRSNN is proposed to improve the security of IoT networks. The developed technique has five steps: preprocessing data, extracting features, equalizing classes, optimizing features, and classification. Normalization and removing outliers are two methods used in data processing to make sure that input data is accurate and consistent. Feature extraction using the RNN part is to find abnormal patterns and high-level features, which are then turned into spike trains for the SNN to process over time. For class equalization, a Synthetic Minority-Oversampling Technique (SMOTE) is used. Recursive Feature Elimination (RFE) is used to keep the important features for feature optimization. Then, the dataset is split into sets for testing and training so that the model can be tested properly. The hybrid model integrates the spatial feature learning skills of RNNs along with the versatility in time of SNNs, leading to improved accuracy and resilience in identifying IoT network abnormalities. The developed HRSNN method works better than the other deep learning (DL) models on the CIC-IoT23 and TON_IoT data sets. Experimental assessments show that the model attained an accuracy rate of 99.5% in the "CICIoT2023" dataset and 98.75% in the "TON_IoT" dataset.