Skip to main content

Invited talk by Ivica Crnkovic

Ivica Crnkovic is a professor in Software Engineering at Mälardalens University in Västerås.

Time: Thu 2013-04-25 11.15 - 12.15

Location: KTH Campus, Department of Machine Design, Brinellvägen 83, Level 2, Room B242

Title: RALF3 -  Software Deployment  for Embedded High Performance Architectures

Abstract: The extensive development of hardware technologies is enabling heterogeneous embedded systems allowing very high input data rates. This makes it possible to process massive amount of data in real-time, which has a huge area of applications in different types of autonomous systems.

Ralf 3 is a SSF (Swedish Foundation for Strategic Research) research program with a objective to provide methods and tools for optimal distribution of software, particular those with heavy calculations and high data rates, on hardware platforms that include Field-Programmable Gate Arrays (FPGAs), Graphics Processing Units (GPUs) and Central Processing Units (CPUs). In particular the objective is to model and analyze the solutions in an early phase of the design, to make it possible to take proper design decision early. This will increase the development efficiency and decrease the development costs, while ensuring high quality solutions.

The overall goal of Ralf 3 is to formulate and implement an appropriate component model for computation- intensive embedded real-time systems with very high data rates, and to provide a support for distribution of software to the heterogeneous platform to achieve optimal results in performance, system response, and other important requirements for embedded systems. In parallel with this work we  are using two cases as proofs as concepts of concepts: a) an underwater robot with stereo cameras and a heterogeneous platform that includes multicore CPUs, FPGA and GPUs, and with ability to recognize objects in real-time under specific conditions with predictable and optimized performance on the heterogeneous platform,  and b) mammography device based on non-harmful microwaves  with ability to build the images based on detected microwaves  scattering with a significant decrease of computation time.

The seminar will give an overview of the challenges and goals of the projects, and some starting work ideas.