Simulation/Co-Simulation Alternatives for Multi-Domain, Multi- Level, Multi-Language Design of Smart Sensor: The SMAC Methodology

Smart Systems represent a broad class of intelligent, miniaturized systemsincorporating functionalities like sensing, actuation and control. Their heterogeneity interms of domains requires handling different levels of abstraction and differentcommunication and synchronization styles, together with different languages andframeworks. This heavily impacts simulation. Directly simulating in a singleenvironment such a degree of heterogeneity is impossible. On the other hand, co-simulation may lead to integration errors and to a loss in terms of accuracy due to theconjunct adoption of heterogeneous domains. Nonetheless, simulation is crucial inthe design and verification loop, as well as to respect budget constraints.This work proposes the SMAC design flow for smart systems. It provides aformalization of the typical abstraction levels and design domains of a smart system.The proposed taxonomy allows identifying a precise role in the design flow for thedifferent simulation and co-simulation scenarios. This will highlight the impact ofheterogeneous and homogeneous models of computation onsimulationperformance.Finally, this works proposes a methodology to move from a co-simulatedheterogeneity to a C++ simulatable homogeneous representation of the entire smartsystem, whose effectiveness is proven on a complex industrial case study.