This section investigates various intrusion detection approaches used to safeguard WSN-IoT networks, wherein the positives and negatives of each model are discussed based on their performance.

Pundir et al. [24] investigated the different types of security challenges in WSN-IoT networks. The different types of security requirements were also discussed in this study for protecting WSN-IoT networks from intrusions. The following categories of potential threats could greatly affect WSN-IoT networks: eavesdropping, impersonation attacks, DoS attacks, malware attacks, database attacks, and man-in-the-middle attacks. Baraneetharan et al. [25] discussed the impacts of using machine learning algorithms for intrusion detection in WSN-IoT systems. In this study, classification, regression, and clustering-based machine learning algorithms were discussed with regard to intrusion detection in WSN-IoT networks. Moreover, the suggested intrusion detection approaches were compared based on the parameters of prediction accuracy, memory requirements, network architecture, and energy consumption. Among other models, the hybrid IDS frameworks are more suitable for WSN due to their improved energy efficiency and precise detection operation. Jiang et al. [26] deployed a lightweight Gradient Boost Mechanism (GBM)-based cyber-physical system for smart-networking environments. Amouri et al. [27] designed a cross-layered IDS-framework-based linear regression model for increasing the security of WSN-IoT networks. The authors aim to detect common malicious activities like blackholes, flooding, and DDoS within networks [28]. The suggested model has the major drawbacks of an increased false-positive rate and time consumption for attack detection.

Singh et al. [29] presented a comprehensive review to examine the different types of machine-learning-based intrusion detection approaches. This paper covers a few well-known and recently developed ML algorithms to highlight their strengths and weaknesses. This will assist researchers in choosing the best algorithm for their studies. Damasevicius et al. [30] utilized a new annotated dataset named LITNET-2020 for classifying normal and intrusive events pertaining to WSN-IoT systems. In addition, the authors suggested some other cyber-attack datasets for IoT security. Safaldin et al. [31] implemented a binary gray-wolf optimization algorithm incorporated with the standard SVM mechanism for detecting intrusions in WSNs. When recommending a fitness function for assessing each subset of the selected feature, the significance of accuracy and the overall number of features were taken into account. According to the total number of features, the prediction performance of the classifier was determined in the cited work. Here, the SVM uses a dimensionality-reduced feature set for intrusion identification and classification. Some of the merits of using SVM include better scalability, high process speed, and low complexity with a reduced feature set.

Krishnan et al. [32] introduced an anomalous intrusion detection and prevention protocol for WSN-IoT networks. The authors aimed to increase the reliability of a network and provide an expanded time frame for an organization. Jayanayudu et al. [33] utilized hybrid Shuffled Frog Leap (SFL) and Ant Lion Optimization (ALO) algorithms to develop an intrusion detection framework for protecting WSN-IoT systems. Typically, securing data while improving energy efficiency is one of the most challenging network problems in present times. Increased attention to security is necessary while monitoring IDS using IoT-WSN systems. The authors of the suggested paper presented a safe routing intrusion prevention architecture for IoT-WSN networks. Moreover, they concentrated on the enhancement of network efficiency and defense against fraudulent attacks. Here, the greedy strategy was used for data routing, offering energy efficient solutions with security. Hussain et al. [34] presented a comprehensive literature review examining various routing strategies for low-powered IoT systems. Here, the assessment was carried out based on identification, screening, eligibility, and inclusion. Moreover, their work investigated the strengths and limitations of several security-based routing methodologies used in WSN-IoT networks. Al Sawafi et al. [35] implemented a hybrid deep-learning-based intrusion detection framework for WSN-IoT networks. In this paper, the authors intended to mitigate security attacks by analyzing a network traffic dataset. According to the pre-trained features, the authors’ framework categorizes normal and malicious networking traffic in the network. Maheswari and Karthika [36] constructed a multi-tiered intrusion detection (MDIT) framework for safeguarding WSN-IoT networks. Here, the Spotted Hyena Optimization (SHO) algorithm, integrated with the standard LSTM deep learning algorithm, was used to detect malicious events in cyber-data.