Shooting Neutrons at Neurons: Radiation Testing of a Spiking Neural Network on Flash-Based FPGAs
This work provides a validated methodology and demonstrates the resilience benefits of on-chip plasticity for neuromorphic processors in radiation-prone environments like space and avionics.
The authors propose and experimentally validate a radiation-testing methodology for neuromorphic processors with on-chip synaptic plasticity, using the ODIN SNN processor on an FPGA exposed to a high-energy neutron beam. They show that enabling Spike-Dependent Synaptic Plasticity (SDSP) significantly extends time to application-level failure and enables partial recovery from bit flips, with modest hardware overhead.
Neuromorphic, or spiking, processors are increasingly being considered for use in harsh, radiation-prone environments such as space and avionics, where energy efficiency and graceful degradation are essential. In this study, we propose and experimentally validate a radiation-testing methodology specifically designed for neuromorphic processors that employ on-chip synaptic plasticity. We map the open-source ODIN SNN processor with Spike-Dependent Synaptic Plasticity (SDSP) onto the FPGA and expose it to a high-energy neutron beam while continuously monitoring MNIST classification accuracy and recording the synaptic state. From these measurements, we extract SEU cross-sections for ODIN's synaptic memory and develop a calibrated fault model to inform a complementary fault-injection campaign. By comparing inference-only and online-learning configurations, we demonstrate that enabling SDSP can significantly extend the time to application-level failure and enable partial recovery from accumulated bit flips, with modest hardware overhead.