Sequencing and Preempting Data Flow Tasks

Abstract

Real-time applications involve the complex sequencing and preemption of different modes of continuous interaction with the environment. This suggests a hybrid language with data flow and sequencing and preemption constructs. To this purpose, we introduce the notion of time interval into the reactive real-time language Signal, which is instant-based, declarative and data-flow oriented. Associating an execution interval with a data flow process specifies a task. Patterns for the sequencing of tasks using time intervals feature parallelism and hierarchical preemption. They are described at the programming language level, as an extension to Signal. They are integrated in the Signal environment by a preprocessor called SignalGTi, and are compatible with its compilation and verification tools.