Semantics-Directed Hardware Generation of Hybrid Systems

Abstract

Cyber-Physical Systems are being used in "safety-critical" domains and have strict timing constraints, meaning that time predictability of the implementations is highly important. In the event that these timing constraints are violated, the correct operation of the system is no longer guaranteed, and serious consequences may occur. Typically, such systems are implemented in software and executed on processors optimised for average-case performance leading to implementations that are difficult, if not impossible, to analyse for their worst case.We propose an approach which is capable of generating hardware implementations of such systems in a semantics-directed manner, ensuring predictability of the final design. Additionally, we present a schema which is used in the description of such Hybrid Systems, captured as Hybrid Input-Output Automata, which is used in this process. Through the example of a grid of cardiac cells and other industrial examples, we show that this approach generates implementations which are almost 15 times faster than an 800MHz ARM Cortex A-9 in the average case, while maintaining timing predictability and having a Worst Case Reaction Time which is over 3,000 times smaller.