High-Level Modeling of Communications in Real-Time Embedded Systems

An embedded system continuously interacts with its environment under strict timing constraints. We model an embedded system using a generalized task graph (GTG). Nodes in a GTG represent individual tasks and edges represent communication between tasks. Tasks are usually concurrent and communicate among themselves to exchange data and perform the functionality of the system. In this paper, we model the communication between the tasks and analyze their impact on the system performance. An interesting and novel aspect of this work is that the communication is modeled using channels nodes in the task graph along with canonical tokens that are user and channel speci c. We explicitly expose the token delivery delay by using channel nodes in the task graph. We introduce the notion of a canonical token for an equivalence class of channels and describe a metric that allows designer to pick implementations for shared channels that satisfy the system's timing requirements. Futher, we propose a re nement methodology that attempts to provide us with better bounds on the delays by using the notion of a process timing simulation of the entire system. In this simulation model, the timing behavior of the individual tasks is the main focus, and the functionality unimportant.