The Internet of Things (IoT) and Cyber-Physical Systems (CPS) works together to make smart homes, healthcare, industries, and critical infrastructure perform better by creating intelligent and adaptable frameworks. These facilitates decentralised decision-making, predictive maintenance and real-time communication nevertheless, it also gives considerable cybersecurity challenges. This study is on comparative examination of five notable security models: Role-Based Access Control (RBAC), machine learning-based Intrusion Detection Systems (IDS), Attribute-Based Encryption (ABE), blockchain trust frameworks, and distributed edge/fog security models. The models are assessed based on critical criteria such as scalability, latency, computational load, adaptiveness, and security robustness. The study shows that Machine Learning -based IDS are very flexible when it comes to finding new and challenging risks, while ABE offers in detail privacy at a considerable expense in terms of power. Blockchain is reliable and trustful, but it has some problems with latency and energy use. On the other hand, RBAC is lightweight and works well in static contexts like industrial Cyber Physical Systems. Edge and fog security has low latency and real-time resilience but it is restricted by the resources. The results show how important it is to use hybrid, layered methods that include lightweight authentication, adaptive detection and decentralised trust. This work adds by comparing security models to CPS layers, trade-offs, and advocating for hybrid frameworks to tackle the diversity of real-world IoT-Cyber Physical System implementations.