Polycyclic aromatic hydrocarbons (PAHs) in urban particulate matter originate from multiple sources, including vehicular emissions, biomass combustion, and petrogenic inputs. Their source identification is essential for designing effective air-quality and sustainable transport strategies; however, conventional diagnostic-ratio approaches are often limited by overlapping source signatures and the uncertainty associated with interpreting individual molecular indicators. This highlights the need for standardized frameworks capable of integrating multiple molecular indicators into a unified source-oriented scheme. This study proposes a semi-quantitative weighted diagnostic ratio matrix that integrates multiple particle-bound PAH ratios and complements multivariate statistical analysis. PM 2.5 and PM 10 were monitored over one year at three sites representing different traffic intensities in an intermediate-sized Latin American city. PAH concentrations were determined and evaluated using Principal Component Analysis (PCA), molecular diagnostic ratios, and the proposed weighted matrix. The matrix integrates diagnostic evidence from multiple PAH ratios into a standardized semi-quantitative scoring scheme for three source categories: biomass combustion, liquid fossil fuel combustion, and petrogenic. PCA identified a dominant combustion-related pattern in both particulate fractions, characterized by high-molecular-weight PAHs and accounting for more than 76% of the total variance in PM 2.5 and 82% in PM 10. The second components were primarily associated with lower-molecular-weight PAHs. The weighted matrix identified biomass combustion as the predominant source across the monitoring sites, followed by liquid fossil fuel combustion, while petrogenic contributions were comparatively minor. Biomass combustion accounted for 54.3%–73.9% of the overall source attribution, indicating that urban PAH levels were influenced not only by traffic-related emissions but also by a broader combustion background potentially associated with residential heating and biomass-burning activities. Agreement between the matrix-based source profiles and PCA patterns provided a consistency check of the source interpretation, while differences among individual diagnostic ratios highlighted the limitations of relying on isolated indicators. The proposed framework provides a transparent, reproducible, and computationally accessible approach for integrating multiple diagnostic ratios into a unified source-oriented assessment. As a screening-level decision-support tool, it may support urban air-quality assessment, emission-source management, and transport-related environmental planning.
Mirabela-Melinda Medvei, Miruna-Mihaela Modiga, Iuliani Aciobăniței, Nicolae Ţăpuş
Privacy-preserving machine learning in Vehicular Ad Hoc Networks (VANETs) must address sensitive on-board data, limited ground-truth annotations, and verifiable model-update integrity. This paper proposes ZK-FL, a decentralized federated learning framework for Intelligent Transportation Systems (ITS) scene classification that combines semi-supervised pseudo-labeling, zk-SNARK-based update validation, and Hyperledger Besu blockchain auditability. Vehicles train local models, quantize the resulting updates, and generate zero-knowledge proofs demonstrating that distributed update probes satisfy a predefined norm-boundedness constraint. Road-Side Units (RSUs) verify the proofs before aggregation and anchor the corresponding audit metadata on-chain. The framework is evaluated on the BDD100K and Mapillary Vistas v2 datasets under multiple non-IID data distributions and five aggregation methods: FedAvg, Norm Clipping, Median, Trimmed Mean, and Krum. Since Mapillary Vistas does not provide scene-level annotations, two scene-label generation strategies, including a semi-supervised pseudo-labeling approach, are introduced. Experimental results demonstrate effective scene classification, verifiable update integrity, and practical proof-generation and verification, supporting the feasibility of privacy-preserving federated learning for ITS.
Dian Liu, Xiaomei Yuan, Zening Zhang, Lambert Zhang, Haoxiang Wang, Yang Liu
Wrong‐way driving on highways is a low‐frequency but high‐risk event that requires reliable detection by future transportation systems and roadside intelligent infrastructure. Existing camera‐based solutions often rely on fixed lane‐direction settings or short‐term trajectory displacement, making them vulnerable to camera viewpoint changes, vehicle occlusion, bounding‐box drift, and stationary vehicles with flashing lights. This study proposes an adaptive calibration and spatio‐temporal filtering framework for robust detection of highway wrong‐way driving under operational surveillance disturbances. The framework first derives lane-level reference directions from road semantic segmentation and historical vehicle trajectories, and then uses lightweight difference hashing to perceive large viewpoint rotations and update direction references online. Candidate wrong‐way trajectories are subsequently verified using a cascaded strategy that integrates temporal continuity filtering, perceptual difference map analysis, and max‐margin adaptive boundary matching. A dedicated highway surveillance dataset containing 99 video sequences was constructed from real-world open‐highway, tunnel, and camera-rotation scenarios, together with a feature-difference dataset for training the secondary verification module. The proposed framework achieved 97.62% precision, 95.35% recall, and a 96.47% F1‐score, while maintaining a low operational false‐alarm rate against representative trajectory‐ and tracking‐based baselines. These results indicate that adaptive infrastructure-side perception can reduce false alarms and support proactive highway traffic safety management in future transportation systems.
Connected and software-defined vehicles now rely on edge infrastructure for access, payment, and identity-linked services. Entry and exit at a parking facility are two of the most frequent points at which a vehicle transacts with that infrastructure, and each carries financial and privacy risk: a plate may be read incorrectly, a payment request may arrive twice, the cloud link may drop, and plate-payment records may stay linkable long after the vehicle has left. We describe a cyber-resilient session protocol built for software-defined vehicular edge infrastructure. The vehicle on-board unit exchanges entry, payment, exit, and closure messages with a roadside edge service over IEEE 802.11p. On the edge side, a short-lived pseudonymous token holds the session together; payment and exit invariants are enforced locally, state is cached while the cloud is unreachable, and injected low-confidence automated license-plate recognition (ALPR) events are routed to a review queue. We modeled the control logic in UPPAAL 5.0.0 as a single-session FogController paired with a receipt deadline observer. The verified properties cover duplicate payment exclusion, penalty exclusion after payment and during dispute, exit-induced session closure, synchronization consistency, deadlock freedom, eventual receipt completion, and absence of receipt deadline violations. A single 11,000-s SUMO/OMNeT++/Veins run then instantiated 200 vehicles across four controlled pre-transmission loss cohorts of 0%, 5%, 10%, and 20%. Of those, 159 sessions became active, 135 payment confirmations and 132 receipts arrived, and 74 sessions completed end to end. Mean application-level response times were roughly 50 ms for token, payment, duplicate payment, and exit responses, and 100 ms for receipts. Injected fault paths comprised 19 uncertain-ALPR events, 8 duplicate payment rejections, 10 fog-cloud outages with a configured 10-s recovery interval, and 5 exit mismatches with a configured 5-s review interval. The two evaluations give complementary formal safety and wireless execution evidence. They also show that one-shot messaging without retransmission limits both completion and token-revocation reliability under loss.
Generative artificial intelligence (GenAI), encompassing large language models (LLMs), multimodal foundation models, diffusion-based traffic simulators, and agentic planners, is entering urban mobility operations across signal control, dynamic routing, demand forecasting, customer service, and autonomous fleet management. This integration has moved ahead of the governance instruments available to transit agencies, planning organizations, and oversight bodies. Conventional transport-AI governance, designed for narrow predictive models with closed deterministic decision envelopes, struggles when applied to generative systems whose outputs are open-ended, context-sensitive, probabilistic, and directly legible to human operators and citizens. This mismatch produces a tractable governance vacuum that cross-sector instruments cannot resolve without transport-specific lifecycle scaffolding. A five-pillar lifecycle governance framework is proposed in this study, comprising transparency and explainability, accountability and liability, equity and non-discrimination, auditability and oversight, and human-in/on-the-loop control. Each pillar is operationalized through an original five-level maturity scale (levels 0–4) anchored in transport failure scenarios. Twelve falsifiable propositions link pillar maturity to observable outcomes (incident severity, mean-time-to-resolution, disparate-impact detection latency, override-exercise rate, regulatory survival, and public trust). Candidate designs include partial least squares structural equation modeling (PLS-SEM), survival analysis, difference-in-differences, and scenario-based public surveys. The framework offers transport authorities an incremental procurement standard (T2/A2/E2/A’-2/H2 baseline), vendors design-phase boundaries, and regulators a lifecycle conformity-assessment scaffold, while supplying researchers with an empirical agenda.
