From Routing Delay Shifts to Silent Data Corruption: Neutron-Induced SEU Effects in AXI

Reporting by ArXiv Systems and ControlRead the original at arxiv.org

Executive Summary

SRAM-based FPGA systems are vulnerable to single-event upsets (SEUs) in configuration memory, which can cause routing resource perturbations and degrade communication fabrics. In Zynq UltraScale+ MPSoCs, these routing disturbances introduce small propagation delay shifts that do not appear logically incorrect but compromise AXI-based data transfers, potentially leading to silent data corruption. This research experimentally investigates the correlation between neutron irradiation effects on routing delays and AXI interconnect failures under neutron exposure. The study utilized a ZCU104 platform with integrated routing delay sensors and an AXI benchmark involving replicated accelerators. Controlled upset emulation was achieved using an internal configuration access port during neutron irradiation experiments. The findings establish a statistical correlation between measured routing delay events and communication failures, allowing for the derivation of cross-sections for both timing shifts and AXI malfunctions. The overall conclusion is that neutron-induced routing perturbations propagate into system-level silent data corruption in UltraScale+ MPSoCs.

Facts Only

* SRAM-based FPGA systems are vulnerable to single-event upsets (SEUs) in configuration memory.
* Routing disturbances can introduce small propagation delay shifts that compromise AXI-based data transfers.
* Neutron irradiation experiments were conducted on a ZCU104 platform integrating routing delay sensors and an AXI interconnect benchmark.
* A frame-level configuration fault injector emulated upsets via the internal configuration access port.
* Measured routing delay events were statistically correlated with communication failures.
* Cross-sections for both timing shifts and AXI malfunctions were derived.
* The experimental results demonstrate how neutron-induced routing perturbations propagate into system-level silent data corruption in UltraScale+ MPSoCs.

Full Take

The investigation establishes a crucial link between physical phenomena (neutron irradiation causing routing delays) and logical failures (AXI communication errors), moving the analysis beyond studying these subsystems in isolation to assessing their joint impact within a complex SoC environment. The novelty lies in establishing an experimental correlation under realistic fault injection, demonstrating that timing shifts are not just theoretical nuisances but tangible precursors to data integrity loss in high-performance systems like Zynq UltraScale+. The pattern recognized here is the amplification of subtle physical effects into catastrophic system failures; the initial uncertainty about whether delay shifts matter is resolved by showing they correlate with observable communication malfunctions. The implication for resilience design is that focusing solely on mitigating SEUs in configuration memory might be insufficient if the resulting timing variations are not monitored and accounted for across the interconnect fabric. This pushes the necessary focus toward cross-layer monitoring—ensuring that physical layer perturbations inform system-level data integrity protocols, which is essential when designing for operation in neutron-rich environments. What subsequent research should focus on is defining thresholds where delay shifts become an actionable indicator versus noise, and developing dynamic mitigation strategies based on these derived cross-sections rather than static fault tolerance mechanisms.

From the original · ArXiv Systems and Control

Electrical Engineering and Systems Science > Systems and Control [Submitted on 24 Sep 2026] Title:From Routing Delay Shifts to Silent Data Corruption: Neutron-Induced SEU Effects in AXI-Based Zynq UltraScale+ MPSoCs View PDFAbstract:SRAM-based FPGA system-on-chip devices are vulnerable to single-event upsets (SEUs) in configuration memory, which may perturb programmable routing resources and…

Read the full story at arxiv.org

Sentinel — provisional

Raw score, 0 to 1 (uncalibrated)0.15

No strong signs of machine writing were found in the source article. Provisional estimate, not a finding that a person wrote it.

The model's own note (unverified)

The text exhibits the highly dense, precise language characteristic of specialized academic or engineering documentation rather than general narrative writing.

This looks only at the wording of the original source article, not at this page's AI-written sections. A small local AI model made this estimate. It has not been checked against known human and machine texts, so treat it as provisional. It cannot show who wrote an article.

What the model noticed in the source article

Sentence length variance is moderate; specialized technical writing exhibits appropriate complexity.

High internal coherence reflecting a focused scientific argument; the structure follows typical research paper presentation.

Standard academic citation and abstract structure observed, indicative of formal submission process.

Content is highly specialized technical prose; no readily apparent signs of LLM-typical rhetorical hedging or generalized fluff.

Signs the model read as against machine writing

Specific referencing of experimental procedures, specific system acronyms (Zynq UltraScale+ MPSoCs, AXI), and the description of custom experimental setup suggest direct engagement with domain-specific research.

Image: arxiv.org · rights & removal