A multilayer deep autoencoder approach for cross layer IoT attack detection using deep learning algorithms

Cybersecurity professionals depend on multi-layered techniques to find even the most minor anomalies that can point to possible attacks given the complexity of network data. Modern threat environment concerns include feature representation, scalability, and flexibility demand for improved techniques. This work presents the Multi-Layer Deep Autoencoder (M-LDAE), especially tailored for cross-layer IoT threat detection, to solve these difficulties. Specifically designed for cross-layer based Internet of Things (IoT) attack detection, the Multi-Layered Deep Auto Encoder (M-LDAE) is introduced in the present research to overcome these challenges. With the use of deep autoencoders hierarchical simplification capabilities, M-LDAE is able to extract latent representations that contain both global and local attributes. This technology effectively safeguards against various cyber threats, including Man-in-the-Middle attacks at the network layer and Distributed Denial of Service (DDoS) attacks at the transport layer of IoT networks. To improve detection and adapt to emerging attack methods, the M-LDAE system employs deep learning algorithms such as RNNs, GNNs, and TCNs. This research proves that M-LDAE can adapt to new attack vectors, enhance detection accuracy, and reduce false positives through extensive simulations, using benchmark datasets and real-world scenarios. A new paradigm for cross-layer based IoT attack detection is presented in this paper, which provides a flexible and robust solution for complete cybersecurity across different IoT domains and thereby improves the field of cyber threat identification.