Two very simple yet opposite challenges that have a big impact on the Internet of Things (IoT) ecosystem's stability and operational performance are energy efficiency and security. Even while data privacy, integrity, and authentication are important, typical security techniques might add extra processing power and battery life requirements, which is not feasible on resource-constrained IoT devices. In an IoT scenario, traditional encryption and key management solutions are inadequate due to limited processing power, memory, and battery capacity. One of the most potent approaches to address the security concerns of the Internet of Things (IoT) in recent years has been hybrid learning approaches, which can make decisions that are adaptable, data-driven, and intelligent. Hybrid learning algorithms can detect anomalies and intrusions while maintaining trust with minimal human intervention. This review paper assesses the relevance and contribution of several hybrid techniques, including Decision Trees, Support Vector Machines (SVM), Random Forests (RF), K-Nearest Neighbors (KNN), and Artificial Neural Networks (ANN), in improving the security of the Internet of Things (IoT). The paper also discusses how hybrid models are becoming increasingly important. These models, such as CNNs and RNNs, can extract more features and detect attacks in real time in large IoT deployments. We pay particular attention to hybrid learning methodologies, which integrate ML/DL models with optimization algorithms to find the optimal balance between detection accuracy and energy consumption. The paper also addresses other important challenges with using ML/DL in heterogeneous IoT contexts with limited resources, including data imbalance, computational complexity, and rapidly adaptive attack patterns. Finally, it identifies gaps in the research and proposes new research directions, such as lightweight language systems, federated learning, and explainable AI-based intrusion detection systems to make the next generation of IoT networks more secure, scalable, and sustainable.
Paper
Full text
Performance Comparison of Hybrid Intrusion Detection Approaches in IoT Systems
Semantic Scholar · 2026
Abstract
Two very simple yet opposite challenges that have a big impact on the Internet of Things (IoT) ecosystem's stability and operational performance are energy efficiency and security. Even while data privacy, integrity, and authentication are important, typical security techniques might add extra processing power and battery life requirements, which is not feasible on resource-constrained IoT devices. In an IoT scenario, traditional encryption and key management solutions are inadequate due to limited processing power, memory, and battery capacity. One of the most potent approaches to address the security concerns of the Internet of Things (IoT) in recent years has been hybrid learning approaches, which can make decisions that are adaptable, data-driven, and intelligent. Hybrid learning algorithms can detect anomalies and intrusions while maintaining trust with minimal human intervention. This review paper assesses the relevance and contribution of several hybrid techniques, including Decision Trees, Support Vector Machines (SVM), Random Forests (RF), K-Nearest Neighbors (KNN), and Artificial Neural Networks (ANN), in improving the security of the Internet of Things (IoT). The paper also discusses how hybrid models are becoming increasingly important. These models, such as CNNs and RNNs, can extract more features and detect attacks in real time in large IoT deployments. We pay particular attention to hybrid learning methodologies, which integrate ML/DL models with optimization algorithms to find the optimal balance between detection accuracy and energy consumption. The paper also addresses other important challenges with using ML/DL in heterogeneous IoT contexts with limited resources, including data imbalance, computational complexity, and rapidly adaptive attack patterns. Finally, it identifies gaps in the research and proposes new research directions, such as lightweight language systems, federated learning, and explainable AI-based intrusion detection systems to make the next generation of IoT networks more secure, scalable, and sustainable.