AN EXHAUSTIVE ANALYSIS OF SEU EFFECTS IN THE SRAM MEMORY OF SOFT PROCESSOR

Abstract

The Embedded system design is characterized by its daily complexity. It integrates a hardware and software parts together on a common platform. These parts may be defective by a spurious signal, subsequently found to be two types of errors. The software and hardware errors can attack the embedded system. In this paper an exhaustive analysis of the effects of Single Event Upset into the Static Random Access Memory occupied area of Aeroflex Gaisler LEON3 processor is presented. It is a soft core pipeline processor that is part of the GRLIB IP library based on Scalable Processor Architecture, SPARC V8,implemented in Virtex-5 FPGA. A new software methodology allowing fault injection is explored and illustrated in order to classify the defective behaviours while executing several benchmarks. This investigation is done by an exhaustive fault injection campaign (More than 200000 transient faults) into SRAM memory of LEON3 considered as a processor. The proposed method makes error rate predictions more accurate compared to other techniques.

Keywords

• LEON3
• SEU
• Software fault injection
• Benchmark
• FPGA