Cybersecurity in IoT Ecosystems Using Lightweight Deep Learning Models

The rapid proliferation of Internet of Things (IoT) devices across diverse sectors has significantly increased the complexity of networked systems, introducing critical cybersecurity challenges. Traditional security mechanisms, reliant on centralized systems and static detection models, often fail to address the dynamic and decentralized nature of IoT environments. This chapter explores the integration of advanced machine learning techniques, particularly Federated Learning, and real-time collaborative threat detection for securing multi-tiered IoT ecosystems. Emphasizing low-latency decision-making, this work highlights the importance of distributed, real-time threat sharing and coordination across IoT devices, edge systems, and cloud infrastructures. By leveraging Federated Learning for decentralized model training, the chapter offers solutions for privacy-preserving, scalable, and adaptive threat detection and response. Moreover, it delves into the optimization of detection models for low-latency applications, ensuring prompt and effective responses to emerging cyber threats. Challenges such as device heterogeneity, data sparsity, and communication overhead are examined, with a focus on practical strategies for addressing these issues. This chapter ultimately provides a comprehensive framework for enhancing IoT security through collaborative, real-time, and decentralized approaches, paving the way for future research in adaptive security architectures. Keywords: IoT Security, Federated Learning, Real-Time Threat Detection, Low-Latency Decision-Making, Distributed Systems, Privacy-Preserving Models.