Intrusion detection in internet of things using improved binary golden jackal optimization algorithm and LSTM

Internet of things (IoT) technology has gained a reputation in recent years due to its ease of use and adaptability. Due to the amount of sensitive and significant data exchanged over the global Internet, intrusion detection is a challenging task in the vast IoT network. A variety of hostile behaviors and attacks are now detected by intrusion detection systems (IDSs), which are difficult or impossible for a single method to identify. An Improved Binary Golden Jackal Optimization (IBGJO) algorithm and Long Short-Term Memory (LSTM) network are used in this paper to develop a new IDS model for IoT networks. Firstly, the GJO is improved by opposition-based learning (OBL). A binary mode of the improved GJO algorithm is used to select features from IDS data in order to determine the best subset selection. IBGJO uses OBL strategy to improve the performance of the GJO and prevents the algorithm from getting trap in local optima by controlling initial population. LSTM is used in the IBGJO-LSTM model to classify samples. Although high detection rates are achieved by machine learning techniques, the efficiency of these methods decreases with the increase in the size of the dataset. To overcome these problems, deep learning methods are more suitable for distinguishing samples from huge amount of data. The proposed model was assessed using the NSL-KDD and CICIDS2017 datasets. For CICIDS2017 and NSL-KDD, the proposed model was 98.21% accurate. The results show that the recognition accuracy of the proposed model is higher than the models BGJO-LSTM, Binary Whale Optimization Algorithm-LSTM (BWOA-LSTM) and Binary Sine Cosine Algorithm-LSTM (BSCA-LSTM). This is likely because the binary mode of the improved GJO algorithm is able to more effectively select the most relevant features from the IDS data and the LSTM is able to more accurately classify the samples. Also, the proposed model has a significantly higher percentage accuracy than Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Naive Bayes (NB).

Automated summary

This study proposes a new intrusion detection system model for IoT networks using an improved Binary Golden Jackal Optimization algorithm (IBGJO) and Long Short-Term Memory (LSTM) network. The proposed model achieves an accuracy rate of 98.21% on the NSL-KDD and CICIDS2017 datasets. The results show that the improved GJO algorithm effectively selects relevant features from IDS data and the LSTM accurately classifies the samples. Additionally, the proposed model significantly outperforms Support Vector Machine, K-Nearest Neighbors, and Naive Bayes.