Abstract:Neural Networks (NNs) are increasingly used in the last decade in several demanding applications, such as object detection and classification, autonomous driving, etc. Among different computing platforms for implementing NNs, FPGAs have multiple advantages due to design flexibility and high performance-to-watt ratio. Moreover, approximation techniques, such as quantization, have been introduced, which reduce the computational and storage requirements, thus enabling the integration of larger NNs into FPGA devices. On the other hand, FPGAs are sensitive to radiation-induced Single Event Upsets (SEUs). In this work, we perform an in-depth reliability analysis in an FPGA-based Binarized Fully Connected Neural Network (BNN) accelerator running a statistical fault injection campaign. The BNN benchmark has been produced by FINN, an open-source framework that provides an end-to-end flow from abstract level to design, making it easy to design customized FPGA NN accelerators, while it also supports various approximation techniques. The campaign includes the injection of faults in the configuration memory of a state-of-the-art Xilinx Ultrascale+ FPGA running the BNN, as well an exhaustive fault injection in the user flip flops. We have analyzed the fault injection results characterizing the SEU vulnerability of the circuit per network layer, per clock cycle, and register. In general, the results show that the BNNs are inherently resilient to soft errors, since a low portion of SEUs in the configuration memory and the flip flops, cause system crashes or misclassification errors.

Failure Probability and Fault Observability of SRAM-FPGA Systems

Fault modeling and characteristics of SRAM-based FPGAs (abstract only).

Cache RAM inductive fault analysis with fab defect modeling

Hardware Versus Software Fault Injection of Modern Undervolted SRAMs

A New Analytical Approach to Evaluate the Radiation Sensitivity of Circuits Implemented on SRAM-Based FPGAs

SRAM-Based FPGA SEU Fault Propagation Model

Modeling the Effect of SEUs on the Configuration Memory of SRAM-FPGA Based CNN Accelerators

Analyzing the Single Event Upset Vulnerability of Binarized Neural Networks on SRAM FPGAs

A Hierarchical Fault Injection System for RISC-V Processors Targeting Single Event Upsets in Flip-Flops

Analysis of the Critical Bits of a RISC-V Processor Implemented in an SRAM-Based FPGA for Space Applications

Study of an Automated Precise SEU Fault Injection Technique

An Accurate Fault Location Method Based on Configuration Bitstream Analysis

Transient Fault Representativenesses Comparison Analysis

An Analytical Delay Model for SRAM-based FPGA Interconnections

A Cycle-Accurate Soft Error Vulnerability Analysis Framework for FPGA-based Designs

Simics-based System Level Fault Injection Platform

Study of Fault-Injection for FPGA Based Fault-Tolerant Computer

Verification and Automatic Error Injection Method of SEM IP Core for SRAM-Based FPGAs

Using Formal Verification to Evaluate Single Event Upsets in a RISC-V Core

Single Event Latch-Up Hardening Using TCAD Simulations in 130 nm and 65 nm Embedded SRAM in Flash-Based FPGAs

Radioresistance Emulation of FPGA Based on Dynamic Partial Reconfiguration