The research, led by Nelson Kodzotse Agbemava of Ghana’s Nuclear Regulatory Authority alongside colleagues at Idaho National Laboratory, examines what happens when Radiation Detection Systems, or RDS, are migrated to cloud environments. These systems rely on Wireless Sensor Networks, distributed arrays of detectors with limited storage, processing power, and battery life, to monitor large geographic areas for abnormal radiation levels. As nuclear power generation expands, including the deployment of Small Modular Reactors, the volume and frequency of radioactive material transport will grow, making wide-area real-time monitoring increasingly essential. Cloud integration offers a compelling answer: rather than maintaining dedicated devices, locations, and software for hosting nuclear data computing tasks, operators need only internet access to leverage scalable computing resources.

Yet the study is careful to note that every organization adopting cloud services must do so deliberately, because without proper security measures, cloud adoption exposes data and introduces an entirely new attack surface. Working with a Cloud Service Provider means surrendering some control over both security and device configuration, a trade-off that carries particular weight in high-consequence nuclear environments where cybersecurity failures can translate directly into physical security threats. The researchers developed a conceptual architecture for a cloud-enabled RDS spanning four layers: a sensor layer of edge devices and IoT detectors, a communication layer using protocols such as LoRaWAN and LTE, a cloud layer handling storage and analytics, and an application layer providing visualization dashboards and decision systems.

To move from architecture to actionable threat analysis, the team applied the STRIDE model, categorizing threats as Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege, and mapped each category against the four architectural layers. Rather than producing an exhaustive enumeration of every theoretically possible threat path, the researchers deliberately scoped their mapping to mechanisms with documented real-world precedent, drawing on case studies including Operation Cloud Hopper, a campaign in which attackers exploited managed service providers to infiltrate multiple organizations simultaneously, and the Triton malware incident, which demonstrated how cyber-physical disruption of industrial control systems is operationally achievable. This produces a smaller but more defensible set of layer-threat pairs grounded in evidence rather than speculation.

Each identified threat was then scored using a qualitative likelihood-impact matrix, a method consistent with IAEA and NIST risk-informed guidance for contexts where empirical incident-frequency data are unavailable. Likelihood was rated Low, Medium, or High depending on whether a directly comparable real-world precedent exists, whether the mechanism is a documented variant of a precedent, or whether it is plausible but not yet observed in a comparable system. Impact was rated based on potential consequence to RDS integrity or availability, with High reserved for threats capable of producing an undetected false radiation reading or a sustained loss of monitoring coverage. The researchers emphasize that this scoring is an explicit, literature-grounded judgment exercise rather than a statistically derived probability estimate.

The results reorder threat priorities in a way that departs from conventional cloud-security thinking. Managed-service-provider and supply-chain compromise, following the Operation Cloud Hopper pattern, emerged as Critical risk, while sensor-data tampering or falsification ranked as High risk, both ahead of the API exploitation and IAM misconfiguration risks that general cloud-security guidance tends to foreground. Communication-layer interception and wireless jamming scored lower, rated Low to Medium. This ordering suggests that for a cloud-enabled RDS specifically, the highest-priority controls are those governing third-party vendor assurance and sensor-data integrity verification rather than the API-hardening and access-control measures most often discussed in cloud-security literature.

Perhaps the study’s most striking analytical contribution is its examination of the asymmetric consequences of data falsification in radiation detection. A falsified or suppressed low reading, a false negative, means an actual elevation in radiological risk goes undetected; in a border-monitoring or facility-perimeter context, this is the more severe failure mode, since it can permit movement of radioactive material or a release event to proceed without triggering containment, evacuation, or interdiction responses. Because no alarm is raised, the failure may not be discovered until a downstream consequence makes it apparent. Conversely, a falsified high reading, a false positive, does not itself create a radiological hazard but can trigger disproportionate operational responses, including evacuation orders, transport interdictions, or diplomatic incidents at border crossings. Repeated false positives also create an incentive for operators to discount genuine alerts over time, a normalization-of-deviance risk that is itself security-relevant.

Because cloud integration concentrates more of the RDS data path, including aggregation, analytics, and visualization, into infrastructure outside the operator’s direct control, the attack surface for inducing either failure mode shifts toward the cloud and communication layers rather than residing solely at the physical sensor. This is the central cyber-physical implication of cloud adoption for radiation detection: a successful compromise need not touch the physical sensor at all to produce a false reading at the decision-maker’s dashboard, a materially different threat model from the field-level manipulation exemplified by Triton. The researchers argue that the risk discussion cannot be reduced to whether the sensor is secure, but must also ask whether the path from sensor to decision-maker preserves data integrity end to end.

In response to these findings, the study proposes a layered cybersecurity framework aligned with IAEA Nuclear Security Series guidance, including NSS No. 33-T on computer security for nuclear facilities, NSS No. 17 on security of nuclear information, and NSS No. 14 on nuclear security recommendations, alongside NIST Cybersecurity Framework and Zero Trust Architecture guidance. The framework comprises five layers: a governance layer establishing policies, compliance requirements, and regulatory alignment; an infrastructure security layer protecting cloud configurations and network architecture; an application security layer securing APIs through authentication, validation, and secure coding practices; a data security layer enforcing encryption, integrity verification, and access control across all data states; and a monitoring and incident response layer providing continuous visibility through logging, SIEM, and threat detection systems. This layered approach ensures that vulnerabilities in one layer do not compromise the entire system, enhancing resilience in high-consequence nuclear environments.

The researchers are explicit about the limits of their contribution. The framework does not introduce new cryptographic primitives, detection algorithms, or technical control mechanisms; existing cloud, nuclear, and OT/ICS security literature already documents the individual controls referenced throughout the paper. The contribution lies instead in the domain-specific synthesis and prioritization of those controls against threat pathways specific to cloud-enabled radiation sensing rather than to cloud computing or industrial control systems in general. The study is conceptual and desk-based, without empirical testing, simulation, penetration testing, or expert validation, and the risk scores reflect structured but qualitative literature-grounded judgment rather than measured incident probabilities. Future work should prioritize expert or Delphi-panel validation of the risk scoring, tabletop exercises modeling the false-positive and false-negative consequence pathways, and prototype implementation of at least the highest-priority controls to move the framework from conceptual alignment toward operational evidence. The researchers conclude that cloud computing may be securely adopted in nuclear sensor networks when supported by robust governance, regulatory alignment, and risk-informed cybersecurity controls, provided that the framework informs rather than substitutes for the expert validation, simulation, and pilot testing needed before operational adoption.

Subject of Research: Cybersecurity risks and a risk-informed protection framework for cloud-enabled nuclear radiation detection sensor networks

Article Title: Securing cloud-enabled radiation detection systems: a risk-informed cybersecurity framework for nuclear sensor networks

Article References: Agbemava, N. K., Mensah, E. A., Appiah, K., & Spirito, C. M. (2026). Securing cloud-enabled radiation detection systems: a risk-informed cybersecurity framework for nuclear sensor networks. Frontiers in Nuclear Engineering, 5, Article 1893691. https://doi.org/10.3389/fnuen.2026.1893691

Image Credits: AI Generated

DOI: 10.3389/fnuen.2026.1893691

Keywords: cloud computing, nuclear security, radiation detection, cybersecurity, sensor networks, STRIDE threat modeling, IAEA, supply chain security, Securing, cloud-enabled, radiation, detection