The rapid evolution of safety-critical cyber-physical systems (CPS), particularly within industrial automation and automotive domains, has intensified the need for architectures that simultaneously guarantee timing predictability, functional safety, and cybersecurity resilience. This paper presents a comprehensive theoretical investigation into the intersection of real-time scheduling, time-sensitive networking (TSN), precision clock synchronization, and fault-tolerant embedded system design. Drawing upon foundational and contemporary literature, the study synthesizes insights from real-time systems theory, component-based software engineering, and emerging networking standards such as IEEE TSN and precision time protocol (PTP). The analysis reveals that while deterministic communication and scheduling frameworks have matured significantly, their integration with robust security mechanisms remains incomplete, especially under adversarial conditions targeting synchronization protocols. Furthermore, the study explores the implications of model-driven architecture (MDA) and component-based design paradigms in enhancing system modularity and certification processes. The methodological approach is qualitative and analytical, relying on cross-referencing established theoretical frameworks and empirical studies to derive architectural principles. The findings indicate that achieving end-to-end resilience requires a co-design approach encompassing hardware redundancy, network determinism, and adaptive security layers. Additionally, emerging automotive zonal architectures and lockstep processing techniques are evaluated as promising directions for achieving fault tolerance in distributed CPS. The discussion highlights key limitations in current standards, including insufficient threat modeling and scalability challenges, and outlines future research avenues such as adaptive scheduling under uncertainty and secure-by-design synchronization mechanisms. This work contributes to the ongoing discourse by providing an integrative perspective that bridges traditionally siloed domains, offering a foundation for the next generation of resilient, secure, and predictable cyber-physical systems.