2 May 2022 in IEEE Internet of Things Journal
The widespread deployment of road sensors in the Internet of Things (IoT) allows for fine-grained data integration, which is a fundamental demand for data-driven applications. Sensing data with inevitable missing and substantial anomalies are unavoidable, due to unstable network communication, faulty sensors, etc. Recent tensor completion studies have demonstrated the superiority of deep learning in imputation tasks by precisely capturing the intricate spatiotemporal dependencies/correlations. However, ignoring the significance of initial interpolation in these methods results in unstable performance, especially for complicated missing scenarios across large-scale data. Additionally, the existing interpolation methods utilize recursive signal propagation along spatiotemporal dimensions, which produce noise accumulation where the dependencies are uncorrelated. In this study, we design a multiattention tensor completion network (MATCN) for modeling multidimensional representation in the presence of missing entries. MATCN sparsely sampled historical fragments and utilized a gated diffusion convolution layer to generate the initial schemes, which mitigate the exposure bias existing in previous traffic imputation models. In addition, we develop a spatial signal propagation module and a temporal self-attention module as the basic stack block of deep networks, which executes representation aggregation and dynamic dependencies extraction at the spatiotemporal level. This architecture empowers MATCN with progressive completion capacities for complex data missing scenarios. Numerical experiments on four real-world traffic data sets with various missing scenarios demonstrate the superiority of MATCN over multiple state-of-the-art imputation baselines.
1 December 2021 in Journal of Transportation Engineering, Part A: Systems
Accurate traffic forecasting on citywide networks is one of the crucial urban data mining applications that accurately provide congestion warning and transportation scheduling. While previous work has made significant efforts to learn traffic temporal dynamics and spatial dependencies, two key limitations exist in current models: (1) most existing approaches solely capture spatial correlations among neighbors on predefined graph structure, and genuine relation may be missing due to the incomplete graph connections; and (2) existing methods are defective to capture the temporal trends because the recurrent and stacking structure employed in these methods suffers from the long-range temporal dependency vanquish problem. To overcome the difficulty in multistep prediction and further capture the dynamic spatial–temporal dependencies of traffic flows, we propose a new traffic speed prediction framework for multiscale graph attention networks (MS-GATNs). In particular, MS-GATNs is a hierarchically structured graph neural architecture that learns not only the local region-wise geographical dependencies but also the spatial semantics from a global perspective. Furthermore, a multiheads attention mechanism is introduced to empower our model with the capability of capturing complex nonstationary temporal dynamics. Experiments on real-world traffic data sets demonstrate that MS-GATNs outperforms the state-of-the-art baselines in long-term forecasting.
26 June 2025 in IEEE Transactions on Intelligent Transportation Systems
Origin-destination flow prediction traditionally depends on intricate numerical simulations of human mobility. Data-driven approaches, including transformers, have disrupted this paradigm with sophisticated predictive models. However, purely data-driven approaches often act as closed box models, failing to reveal the underlying mechanisms of human mobility. We address these limitations with physics-informed mobility perception networks (PI-MPN), a continuous-time process that implements a key principle of diffusion from human mobility, and introduces the principle into data-driven deep learning networks. PI-MPN consists of data-driven mobility perception networks and physics-informed neural diffusion networks PI-MPN models precise spatio-temporal movement by data-driven networks, learning urban human mobility as a neural flow by physics-informed networks. We conduct comprehensive evaluations on two real-world traffic datasets, revealing the superiority of PI-MPN over existing data-driven benchmarks. Moreover, case analyses demonstrate that the diffusion process can describe the underlying urban dynamics, thus enhancing the model interpretability of human mobility. Model implementation is available at https://github.com/xuesong-wu/PI-MPN
6 July 2024 in Information Sciences
12 October 2022 in IEEE Intelligent Transportation Systems Magazine
The growing use of probe vehicles generates a huge number of global navigation satellite systems (GNSS) data. Limited by satellite positioning technology, further improving the accuracy of map matching (MM) is challenging work, especially for low-frequency trajectories. When matching a trajectory, the ego vehicle’s spatial-temporal information of the present trip is most useful with the least amount of data. In addition, there is a large number of other data, e.g., other vehicles’ state and past prediction results, but it is hard to extract useful information for matching maps and inferring paths. Most of the MM studies have used only the ego vehicle’s data and ignored other vehicles’ data. Based on those, this article designs a new MM method to make full use of “big data.” We first sort all the data into four groups according to their spatial and temporal distance from the present matching probe, which allows us to sort for their usefulness. Then we design three different methods to extract valuable information (scores) from them: a score for speed and bearing, one for historical usage, and another for traffic state using a spectral graph Markov neural network. Finally, we use a modified top-K shortest-path method to search the candidate paths within an ellipse region and then use the fused score to infer the path (projected location). We test the proposed method against baseline algorithms using a real-world dataset in China. The results show that all scoring methods can enhance MM accuracy. Furthermore, our method outperforms the others, especially when the GNSS probing frequency is ≤ 0.01Hz.
13 May 2026 in SSRN Electronic Journal
The increasing deployment of sensor-equipped inspection agents and digital asset management systems has made city-wide road inspection larger and more dynamic. Learning-based coordination methods, particularly reinforcement learning, provide a principled way to address this routing complexity. However, existing studies often remain centered on static coverage-oriented routing, with limited consideration of risk exposure and the communication or privacy constraints that shape coordination in practice. To this end, we reformulate urban road inspection as a multi-agent routing task under hierarchical coordination regimes to reduce network risk exposure. First, we design a heterogeneous coordination paradigm that reflects differences in information availability and centralized scheduling. Second, we develop a multi-agent coordinated routing policy (MACRP) to enable effective agent collaboration during parallel solution construction. Its communication module regulates information exchange, enabling coordinated decisions across regimes. Finally, experiments on real-world inspection records show that our framework achieves more consistent and effective reductions in system risk than traditional and neural baselines. Further analyses reveal how regime penetration affects performance and how routing quality degrades under restricted coordination.