Architecture and design choices for an AI-enabled FPGA-based cosmic radiation sensor

Abstract

Field Programmable Gate Array (FPGA) devices can be reconfigured after their manufacture to accelerate tasks such as Artificial Intelligence (AI) inference, provide lower response times and in some cases even lower power consumption compared to the software equivalent running on a modern microprocessor due to the capability of accessing their resources in parallel. Complex physics experiments and Space applications benefit greatly from FPGAs. We propose using a general-purpose, low-power, low-cost FPGA device in a nanosatellite as a cosmic radiation sensor. The AI-based hardware Cosmic Radiation Sensor (AICoRS) will be able to detect single event upsets (SEUs) and analyze the radiation patterns on the fly. The AICoRS system will be radiation tolerant to ensure correct operation in space conditions while acting as a SEU detector. We present the proposed architecture and the design choices we face, as well as a preliminary study of implementing the SEU detection element for evaluating the feasibility of using such a device for Space applications and for measuring cosmic radiation from the point of view of power consumption.