Cyber-Physical Power Systems (CPPS) increasingly inherit cybersecurity vulnerabilities from industrial control components, yet practitioners lack a CPPS-fo
We present a hybrid framework for adaptive insider-threat detection that tightly integrates multi-agent simulation (MAS), layered Security Information and Event Management (SIEM) correlation, behavioral and communication forensics, trust-aware machine learning, and Theory-of-Mind (ToM) reasoning. Intelligent agents operate in a simulated enterprise environment, generating both behavioral events and cognitive intent signals that are ingested by a centralized SIEM. We evaluate four system variants: a Layered SIEM-Core (LSC) baseline, a Cognitive-Enriched SIEM (CE-SIEM) incorporating ToM and communication forensics, an Evidence-Gated SIEM (EG-SIEM) introducing precision-focused validation mechanisms, and an Enron-enabled EG-SIEM (EG-SIEM-Enron) that augments evidence gating with a pretrained email forensics module calibrated on Enron corpora. Across ten simulation runs involving eight malicious insiders, CE-SIEM achieves perfect recall (1.000) and improves actor-level F1 from 0.521 (LSC) to 0.774. EG-SIEM raises actor-level F1 to 0.922 and confirmed-alert precision to 0.997 while reducing false positives to 0.2 per run. EG-SIEM-Enron preserves high precision (1.000 confirmed-alert precision; 0.0 false positives per run), slightly improves actor-level F1 to 0.933, and reduces detection latency (average TTD 10.26 steps versus 15.20 for EG-SIEM). These results demonstrate that cognitive context improves sensitivity, evidence-gated validation enables high-precision, low-noise detection, and pretrained communication calibration can further accelerate high-confidence insider threat identification.
Autonomous clinical agents increasingly act on Internet-of-Medical-Things (IoMT) endpoints, yet existing access-management mechanisms are human-centric, quantum-vulnerable, and poorly suited to continuous action-level enforcement. We present PQ-MedTrust, which couples an identity–behavior–context trust score with risk-dependent decay to post-quantum revocation of agent credentials based on Decentralized Identifiers and Verifiable Credentials, and we characterize where behavioral trust succeeds, becomes detection-latency limited, or fails. In a liboqs testbed on a general-purpose CPU, classical, hybrid, and PQC-only handshakes stay below 1.5 ms, but a hybrid authorized action (handshake, credential, delegation token, and request) occupies 21.1 KB against 1.8 KB classically. The trust-to-revocation path, which signs and pushes a revocation record, takes 0.07 ms with classical and 11.1 ms with SLH-DSA signatures. Over 1000 simulated runs per scenario, policy enforcement revokes every out-of-scope and one-shot mimicry attack (scope is a generator label, not a dose-range check), whereas adaptive stealth is detected in 20.2% of runs at 24 h and 75.3% at 120 h; without the generator’s risk labels, 24 h detection falls to 5.6%. The pooled benign false-positive rate is 1.8% under a reliable context but rises sharply when context degrades or benign agents emit the event class that marks adaptive stealth. A graduated response revokes none of 500 unannounced-emergency runs, although 92.4% reach clinician confirmation, and still revokes every policy violation.
Cyber-Physical Power Systems (CPPS) face a rapidly growing number of cybersecurity vulnerabilities, yet no structured dataset or unified risk scoring method exists to prioritize these CVEs. This paper presents a cohesive methodology for collecting, enriching, and modeling CPPS-related CVEs to predict their risk and prioritize remediation. We aggregate over 4,030 ICS-relevant CVEs from public sources (2020--2025) and enrich each with industry-standard severity (CVSS), exploitation data (CISA Known Exploited Vulnerabilities, Exploit Prediction Scoring System), and contextual ICS attributes. We then develop a two-stage machine learning pipeline using gradient-boosted decision trees and linear regression models to estimate a ''CPPS risk score'' for each CVE and the probability of exploitation, leveraging both structured features and CVE textual descriptions. The novelty of our work lies in building one of the first structured CPPS vulnerability datasets and integrating OT-specific context with threat intelligence for risk prediction, bridging a gap not covered by existing severity metrics alone. Our results show that the proposed models achieve high predictive performance (up to R 2 =0.93 for risk regression and PR--AUC ≈ 0.98 with Brier score ≈ 0.0018 for exploit prediction), enabling accurate ranking of vulnerabilities. Applied to the CPPS dataset, the framework concentrates attention on roughly 30% of CVEs classified as Extreme or High priority while demoting about half of the vulnerabilities to Low priority, and it surfaces several high-risk issues in critical products such as protection relays and SCADA servers that are not yet listed in CISA's KEV catalog. The curated dataset and risk scoring pipeline provide a reusable foundation for data-driven vulnerability management in power systems.
Autonomous clinical agents increasingly act on Internet-of-Medical-Things (IoMT) end points, yet existing identity and access-management mechanisms remain human-centric, quantum-vulnerable, and poorly suited to continuous action-level trust enforcement. We present PQ-MedTrust, a three-tier post-quantum trust-to-revocation architecture integrating Decentralized Identifiers, Verifiable Credentials, identity–behavior–context trust with risk dependent decay, hybrid classical/post-quantum authentication, and automatic credential revocation. We implement and measure the complete enforcement path and characterize where behavioral trust succeeds, becomes detection-latency limited, or fails to separate malicious from benign behavior. A liboqs-based testbed evaluates classical-only, hybrid, PQC-only, and SLH-DSA configurations. On a general-purpose CPU, classical, hybrid, and PQC-only handshakes remain below 1.5ms, while hybrid authorized actions expand to 21.1KB versus 1.8KB classically. Cryptographic enforcement adds 0.07–11.1ms. Across 1,000 runs per scenario, aggressive scope abuse and one-shot mimicry are detected in 100% of runs, while adaptive stealth reaches 20.2% detection at 24h and 75.3% at 120h. The pooled benign false-positive rate is 1.8% under reliable context, but rises sharply when context is degraded or benign agents emit the unusual-but-in-scope behavior used to characterize adaptive stealth. These results show modest CPU-tier PQC cost but substantial communication overhead, while exposing the operating limits of automatic behavior-driven revocation.