Efficient Intrusion Detection System In Iot Using Hybrid Deep Learning Algorithm

Abstract

The Internet of Things (IoT) has become integral to numerous applications, yet it continues to face significant
security challenges despite the introduction of various protective measures. To address these vulnerabilities,
this paper proposes a Hybrid Deep Intrusion Detection System (HDIDS) designed to enhance security in IoT
environments. The proposed system combines Spiking Neural Networks (SNN) and the Lion Optimization
Algorithm (LOA) to effectively detect intrusions. The process begins with the preprocessing of raw data,
followed by feature extraction. In the classification phase, an SNN is utilized to categorize the data as either
normal or indicative of an attack. The classification accuracy is further optimized through hyper parameter
tuning of the SNN using LOA. The performance of the proposed HDIDS is evaluated using the KDD99 dataset.
Experimental results demonstrate that the proposed SNN-based intrusion detection system outperforms existing
methods in terms of key evaluation metrics, confirming its effectiveness in securing IoT environments.