Vehicular systems are increasingly controlled by distributed software components that demand a reliable and secure communication infrastructure. Ethernet and Time-Sensitive Networking (TSN) offer a promising foundation for future high-bandwidth, real-time capable In-Vehicle Networks (IVNs). Current IVNs, however, are vulnerable to safety-compromising attacks, calling for a multifaceted security hardening. In this paper, we present a control and monitoring framework that guards safety and security throughout an attack-resistant real-time network architecture with effective misbehavior detection. Software-Defined Networking (SDN) control of TSN networks enables the enforcement of security policies and the dynamic adaptation to changing conditions. A combination of network monitoring techniques, traffic analysis, observable metrics, and detection algorithms tailored for automotive deployment enable the detection of anomalies and attacks. This network-centric approach enhances the security of existing automotive protocols and Electronic Control Units (ECUs) of limited computing power and remains extensible for future demands. Evaluation results from simulations, a hardware test bed, and a real prototypic vehicle demonstrate the feasibility, effectiveness, and remaining challenges of our approach.