Cyber-physical systems (CPS) are engineered systems that are characterized by the non-trivial interaction of computational components with physical processes. In order to ensure the safety and reliability of such systems, a multitude of approaches have been explored that aim to formally verify that the CPS is guaranteed to behave as intended. However, these approaches often fail to scale to complex systems or are inapplicable in certain cases, e.g., when no accurate model of the system is available. In this thesis, we focus on a complementary approach, called online monitoring. It involves the real-time observation of the evolution of a CPS in order to detect safety violations and potentially trigger alerts and corrective actions. We develop a flexible and expressive formalism for specifying quantitative properties of CPS and online monitors for these properties. Our formalism can be viewed as a domain-specific language (DSL) that describes signal transformations. A key feature of our DSL is that it relaxes the causality restriction of similar prior approaches, by allowing the output to depend on a bounded amount of future input. We illustrate the usefulness of our DSL by using it (1) to implement an ECG monitoring application, and (2) to encode online monitors for quantitative temporal properties.