Athens University Economics and Business (AUEB)
Managing information flow risk propagation in critical IIoT infrastructures
Abstract
dc:descriptionIndustrial Internet of Things (IIoT) systems—such as smart grids—run critical processes where many devices, services, and applications exchange data across layers. Traditional access control decides who may start an action, but it does not track where the data flows next. In dynamic, multi-hop environments, this gap can enhance unauthorised access to critical information. This dissertation tackles that gap with an information flow control (IFC) approach that models real business processes as a directed graph and updates risk in real-time as new transactions occur. The goal is to help operators identify risky paths early and acton them with targeted policies.We define six objectives that guide our work: build a precise graph model of processes and flows (O1); estimate risk along multi-hop paths, not just single connections (O2); update risk after each transaction to reflect new evidence (O3); identify influential nodes using risk-aware centrality, not structure alone (O4); turn analytics into concrete policy changes (O5); and set fair baselines for evaluation (O6). The core contribution is INFFLOW-RT, areal-time, adaptive methodology for risk-based IFC in IIoT. First, we enrich the dataset by adding all plausible transactions between objects, so indirect dependencies are visible. Second, we assign transaction-level weights (operation type and legality) and combine them with severity to estimate impact at edge, node, and path level. Third, we apply incremental Bayesian updates so risk changes with every new event, without recomputing the whole graph. Finally, we stabilize estimates under sparse data using Laplace smoothing. Together, these steps support dynamic, context-aware detection of high-risk flows and the nodes that most of ten involved in them. We evaluate the approach on eight business processes drawn from real smart-grid scenarios. We first build a process-oriented dependency graph that captures legal flows and their transactions (e.g., bulk meter readings entering an AMI head-end, then moving to MDM, CIS, and downstream portals). We show how graph-based enrichment increases coverage, then compute several supporting metrics (severity, operation factor, legality) that drive to the calculation of transaction impact and scaled impact, and finally obtain static and dynamic risk estimates. Because dynamic label classification was out of our scope, we focus on demonstrating timely risk convergence, early detection, and stable prioritization rather than classifiers cores. The results highlight how dynamic updates and risk-weighted centrality identify high risk nodes and paths earlier than static analyses and structure-only baselines.Turning analysis into action is essential. We propose a policy feedback loop that converts high-risk alerts into small, testable policy changes (e.g., blocking or rate-limiting certain operations on specific paths, adding sanitization at control points, or tightening data recipients), with quick “what-if” checks against recent traffic before enforcement. This closes the loop from detection to mitigation and helps reduce illegal flows without disrupting normal business operations. We also explore how to enforce decisions and keep reliable audit trails at scale. A supporting study examines combining capability-based access control (CapBAC) with blockchain to issue, check, and revoke fine-grained permissions, while recording evidence in an immutable ledger. We discuss a practical hybrid design: keep authorization state and audit on a permissioned chain for integrity and traceability, and perform high-rate enforcement at gateways/PLCs with fast revocation checks pushed from the chain. This balances security, speed, and accountability in industrial settings.Finally, to support realistic evaluation, we also surveyed industrial emulation/simulation tools used to bring plant scenarios into the lab. This review helped shape requirements for building test scenarios, assessing tooling coverage (e.g., protocol support, timing fidelity), and understanding where to prototype enforcement.
Degree
thesis:*- Grantor dc:publisher
- Athens University Economics and Business (AUEB)
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Anagnostopoulou, Argiro-Angeliki
- Αναγνωστοπούλου, Αργυρώ-Αγγελική
Subjects
dc:subject × 12- Ελεγχος ροών πληροφορίας
- Έλεγχος πρόσβασης
- Προστασία κρίσιμων υποδομών
- Information flow control
- Access control
- Critical infrastructure protection
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Επιστήμες Μηχανικού και Τεχνολογία
- Υπολογιστές, Υλικό (hardware) και Αρχιτεκτονική
- Electrical Engineering, Electronic Engineering, Information Engineering
- Engineering and Technology
- Computer science, Hardware and Architecture
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/61438