Mixed Criticality Systems (MCS), very popular nowadays, require the simultaneous execution of several environments with different needs in terms of real-time or security level. To achieve this, heterogeneous embedded systems are used, where, through virtualization based on hypervisors or AMP (Asymmetric Multi-Processing) frameworks, several environments with different operating systems can be executed on a single device. However, this is a challenge in terms of compliance with real-time and isolation. The thesis focuses on these two characteristics, applied on System-on-Chip devices with a Field-Programmable Gate Array (SoC FPGA). On the one hand, different latencies introduced by various virtualization technologies are measured, and a circuit is designed that facilitates temporal analysis, which represents an advance with respect to existing tools. On the other hand, a method is implemented to extend the logical contexts provided by hypervisors to the FPGA, improving system isolation and security.