Applying RMA to improve a high-speed, real time data acquisition system

Abstract

Unlike other real-time systems, data acquisition applications spend most of their time performing synchronous and asynchronous I/O operations. This fact, that affects the tasks' abilities to meet deadlines, can be modelled with some extensions to the classic rate monotonic schedulability test. In this paper, we address this issue by investigating a case study for the application of rate monotonic analysis (RMA) to data acquisition systems. We describe a technique for modelling tasks that perform asynchronous I/O at the end of their executing period; and we also show the logistic and technical problems that can arise when applying RMA at different stages during the development process of an application. We illustrate our ideas with a real-time data acquisition system that we have developed, characterized by a very heterogeneous architecture whose technological components have been pushed to the limits of their performance and capacity.<>