Migration of Produkt from ARM9-based to FPGA-based Architecture: HW/SW-partitioning and analysis of Architecture

Abstract

The origins of this thesis is a project currently in development at Laerdal Medical A/S, a company developing first aid equipment and training solutions for medical personnel. One of their most prestigious projects it SAPS, an advanced patient simulator giving students the opportunity to practice a wide range of medical procedures. Internally the patient simulator contains an advanced embedded system with a 32-bit microcontroller and an FPGA working in unison to handle all the data processing required by the patient simulator. The engineers at Laerdal wanted to see if the ARM9 microcontroller could be removed from the system entirely. Modern FPGAs are increasing in power and several microcontroller architectures are available for implementation on FPGAs. The goal of this thesis is to examine the possibility of replacing the microcontroller in the patient simulator with one or more MicroBlaze microcontrollers implemented on the FPGA, as this would reduce system cost and complexity. An architecture is proposed to take over the tasks from the original microcontroller and a simple test software is implemented to show whether or not the architecture has the necessary performance to implement the original software. Some problems with the practical implementation result in only a simplified test case being performed, and this shows very poor performance compared to that of the original system. As such the proposed architecture is not recommended for implementation in the SAPS project, but some thoughts on an eventual implementation in a new project are given and the problems discovered with the tests are explained.