Multi-tiered Artificial Neural Networks model for intrusion detection in smart homes

In recent years cybersecurity has become a major concern in the adaptation of smart applications. A secure and trusted mechanism can provide peace of mind for users, especially in smart homes where a large number of IoT devices are used. Artificial Neural Networks (ANN) provide promising results for detecting any security attacks on smart applications. However, due to the complex nature of the model used for this technique, it is not easy for common users to trust ANN-based security solutions. Also, the selection of the right hyperparameters for ANN architecture plays a crucial role in the accurate detection of security attacks. This paper proposes Multi-tiered ANN Model for Intrusion Detection (MAMID) that is a novel and scalable solution for optimal hyperparameter selection to detect security attacks with high accuracy. The explainability analysis of the predictions made by the model is also provided for establishing trust among users. The approach considers a subset of the dataset for quick, scalable and optimal selection of hyperparameters to reduce the overhead of the process of ANN architecture design. Using a very recent IoT dataset the proposed approach showed high performance for intrusion detection with 99.9%, 99.7%, and 97.7% accuracy for binary, category, and subcategory classification of attacks. To the best of the authors' knowledge, no previous research work has been able to achieve attack detection at the subcategory level beyond 90%.