Detection and Protection of a COTS System Against Micro-Single-Event-Latchups (μ-SELs) and SELs: A Step Toward the Next-Paradigm in New Space

We envision that the “next-paradigm” of New Space will involve the employment of commercial-off-the-shelf (COTS) systems (embodying multiple COTS ICs) vis-à-vis individual COTS ICs today. Most COTS ICs and COTS systems are susceptible to radiation effects, where micro-single-event-latchups (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu $ </tex-math></inline-formula>-SELs) and single-event-latchups (SELs) are of the most concern. Since reported approaches are limited to protecting an individual COTS IC, their application to every COTS IC within the system would render considerable overheads and potentially degraded performance. In this article, we propose an approach to detect and protect a COTS system (vis-à-vis an individual COTS IC) against <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu $ </tex-math></inline-formula>-SELs/SELs, making a step toward our envisioned “next-paradigm” of New Space. Our proposed approach monitors only the current of the COTS system’s total power rail, where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu $ </tex-math></inline-formula>-SELs/SELs may occur in any COTS IC therein. Of note, as this current is obtained outside of the COTS system, the COTS system is unmodified. We propose to adopt the long short-term memory (LSTM) neural network for <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu $ </tex-math></inline-formula>-SEL/SEL detection. Our proposed artificial intelligence (AI)-based approach achieves high (~95%) detection accuracy with fast (~0.5 ms) response time and low (~73 mW) power dissipation. Specifically, our approach improves accuracy by ~15%–60% over reported approaches and matches our prior work on an individual COTS IC (not a COTS system). It reduces power dissipation by ~11%–64% compared to reported approaches and is ~25%–55% higher than our prior work; its response time is <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1.6\times $ </tex-math></inline-formula>–<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$255.6\times $ </tex-math></inline-formula> faster than reported approaches and is <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.2\times $ </tex-math></inline-formula>–<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.4\times $ </tex-math></inline-formula> slower than our prior work. In view of these attributes, our proposed AI-based approach is appropriate for the “next-paradigm” of New Space.