Open research questions in Traffic Prediction and Management Techniques

What the literature leaves open

• Existing methods face challenges in capturing complex spatiotemporal dependencies among road segments. Computational efficiency is a challenge in large-scale urban networks. Prior work has limitations in terms of grid resolution and computational cost.

  An enhanced dynamic spatiotemporal residual network with multi-scale gridding for network-scale traffic speed prediction · 2026 · DOI

• Exploring how federated learning is applied in the field of traffic flow forecasting. Improving model prediction performance and efficiency in federated learning.

  A Short-Term Traffic Flow Prediction Method Based on Personalized Lightweight Federated Learning · 2025 · DOI

• The low quality of traffic data limits the improvement of prediction accuracy. Existing models are overly bulky and redundant, leading to high computational complexity. The lack of personalization in federated learning models can result in poor performance for clients with diverse needs.

  A Short-Term Traffic Flow Prediction Method Based on Personalized Lightweight Federated Learning · 2025 · DOI

• Expanding the building sample to include more diverse architectural styles, cultural contexts, and climate regions, - Implementing continuous learning mechanisms to allow the model to adapt to changing patterns over time, - Capturing the full range of long-term trends and cyclical patterns in building energy consumption

  Deep Learning Framework Using Transformer Networks for Multi Building Energy Consumption Prediction in Smart Cities · 2025 · DOI

• Existing models' limited ability to capture cross-building correlations. Existing models' poor scalability when deployed at urban scale. Inadequate integration of diverse data streams in existing models.

  Deep Learning Framework Using Transformer Networks for Multi Building Energy Consumption Prediction in Smart Cities · 2025 · DOI

• Recent time-series foundation models (FMs) report strong zero-shot accuracy on heterogeneous forecasting benchmarks, but it remains unclear whether these gains transfer reliably to pedestrian sensing deployments.

  How Good Are Time-Series Foundation Models for Pedestrian Crowd Count Forecasting? A Cross-Dataset Comparative Study · 2026

• Centralized forecasting is frequently obstructed by heterogeneous cross-city data and stringent data privacy regulations. Traditional time series forecasting methodologies are insufficient for capturing complex spatio-temporal patterns.

  Deep heterogeneity learning for cross-city transit forecasting: a differentially private federated framework with mixture-of-experts and seasonal decomposition · 2026 · DOI

• Future research will explore adaptive DP mechanisms that dynamically adjust noise levels based on data sensitivity, model convergence, or specific client contributions, aiming for greater efficiency and tighter privacy guarantees with minimal accuracy sacrifice. Future work could focus on developing XAI techniques specific to federated MoE models to understand how different experts contribute to predictions for specific cities or scenarios, and how DP impacts interpretability. • Exploring Advanced Federated Personalization Strategies: While MoE provides adaptation, more explicit personalized federated learning (pFL) techniques, such as FedMeta, FedAvgM, or client clustering, could be investigated.

  Deep heterogeneity learning for cross-city transit forecasting: a differentially private federated framework with mixture-of-experts and seasonal decomposition · 2026 · DOI

• The study identifies a gap in the current traffic congestion prediction methods, which can be addressed by using hybrid predictive models and optimization methods. The paper highlights the need for more efficient data-driven transportation management systems.

  Multimodal traffic flow analysis and congestion prediction on expressways: evaluating the importance of machine learning models and features for improving prediction accuracy in urban traffic management · 2026 · DOI

• The paper does not evaluate how model performance degrades under specific real-world traffic conditions (e.g., rush hours vs. off-peak periods, adverse weather events, accidents, special events) or how the trained models transfer across different expressways with different traffic characteristics.

  Multimodal traffic flow analysis and congestion prediction on expressways: evaluating the importance of machine learning models and features for improving prediction accuracy in urban traffic management · 2026 · DOI

• Requires high computational power - Accuracy drops in very dense crowds - Affected by weather and lighting conditions - Needs large datasets for training

  Public Transport Crowd Estimation via Bus Stop Images Using AI · 2026 · DOI

• Investigating the system's performance in different environments and scenarios. Improving the system's robustness to noise and interference. Integrating the system with other modalities, such as radar or lidar.

  False Positive Reduction in Emergency Vehicle Detection Using a Multimodal Edge-Based System · 2026 · DOI

• Traditional vision-based emergency vehicle detection systems suffer from high false positive classifications. Prior systems are prone to limitations, including the inability to distinguish between active and inactive emergency vehicles.

  False Positive Reduction in Emergency Vehicle Detection Using a Multimodal Edge-Based System · 2026 · DOI

• Theoretical analysis of the convergence of the proposed method. Application of the proposed method to real-world problems. Extension of the proposed method to other optimal transport problems.

  A deep learning approach to multi-marginal optimal transport via Hilbert space embeddings of probability measures · 2026 · DOI

• The lack of a computationally efficient solution for large-scale multi-marginal Monge problems. The need for a novel numerical method that can handle multiple target distributions.

  A deep learning approach to multi-marginal optimal transport via Hilbert space embeddings of probability measures · 2026 · DOI

• To apply the proposed approach to other types of data, such as non-high-frequency datasets. To explore the use of other machine learning algorithms for parameter estimation in the FPP. To investigate the application of the proposed approach to other fields, such as biology and engineering.

  NeuroMem-FPP: A recurrent neural approach for memory-aware parameter estimation in fractional Poisson process · 2026 · DOI

• The classical Poisson model cannot capture complex temporal patterns in real-world systems. Traditional methods such as the method of moments have limitations in estimating the parameters of the FPP. There is a need for a more accurate and efficient approach for parameter estimation in complex fractional temporal point processes.

  NeuroMem-FPP: A recurrent neural approach for memory-aware parameter estimation in fractional Poisson process · 2026 · DOI

• Data scarcity in traffic networks. Estimating a realistic MFD with a limited number of loop detectors. Transferring knowledge from data-rich cities to cities with scarce data.

  Learning to learn the macroscopic fundamental diagram using physics-informed and model agnostic machine learning · 2026 · DOI

• The model is trained and tested on a specific dataset (UTD19) - The approach is limited to estimating the MFD for cities with scarce data from loop detectors - The model's performance may be affected by the spatial distribution of LDs

  Learning to learn the macroscopic fundamental diagram using physics-informed and model agnostic machine learning · 2026 · DOI

• Data imbalances pose challenges, particularly when working with a limited number of datasets. Road structures can affect the accuracy of accident detection.

  Smart City Transportation Deep Learning Ensemble Approach forTraffic Accident Detection · 2026 · DOI

• Incorporating more sophisticated deep learning algorithms and hybrid models can improve detection accuracy. Adopting edge computing can reduce latency and enable faster real-time decision-making. Integrating emerging technologies, such as 5G communication and advanced IoT frameworks, can enhance data transmission speed and system scalability.

  Smart City Transportation Deep Learning Ensemble Approach forTraffic Accident Detection · 2026 · DOI

• To explore the use of other machine learning models for spatiotemporal traffic forecasting. To evaluate the framework on more datasets and in different scenarios. To improve the computational efficiency of the framework.

  DG-LLM: Decomposition-based dynamic graph adaptation of large language models for spatiotemporal traffic forecasting · 2026 · DOI

• Existing methods for traffic forecasting struggle with modeling complicated spatiotemporal dependencies and capturing long-term patterns. There is a need for a novel approach to spatiotemporal traffic forecasting that can effectively model dynamic spatial dependencies.

  DG-LLM: Decomposition-based dynamic graph adaptation of large language models for spatiotemporal traffic forecasting · 2026 · DOI

• Rapidly changing vehicular mobility and demand patterns. Limited effectiveness of existing content prediction and caching strategies. The need for adaptive cache decision-making in dynamic IoV environments.

  A Hybrid Deep Learning Architecture for Content Request Prediction in the Internet of Vehicles · 2026 · DOI

• Large-scale experimentation to validate the proposed approach. Exploration of other deep learning architectures for content request prediction. Investigation of the impact of varying demand distributions on the proposed approach.

  A Hybrid Deep Learning Architecture for Content Request Prediction in the Internet of Vehicles · 2026 · DOI

Most-cited papers in Traffic Prediction and Management Techniques

• Extraordinary Rendition and the Constitution: The Case of Maher Arar · ˜The œReview of litigation · 2008 · 557 citations
• Predicting citywide crowd flows using deep spatio-temporal residual networks · Artificial Intelligence · 2018 · 478 citations
• Real-time crash risk prediction on arterials based on LSTM-CNN · Accident Analysis & Prevention · 2019 · 362 citations
• Graph Neural Networks for Intelligent Transportation Systems: A Survey · IEEE Transactions on Intelligent Transportation Systems · 2023 · 335 citations
• Sensing Data Supported Traffic Flow Prediction via Denoising Schemes and ANN: A Comparison · IEEE Sensors Journal · 2020 · 269 citations
• A spatiotemporal deep learning approach for citywide short-term crash risk prediction with multi-source data · Accident Analysis & Prevention · 2018 · 264 citations
• A Decomposition Dynamic graph convolutional recurrent network for traffic forecasting · Pattern Recognition · 2023 · 256 citations
• Long-term traffic flow forecasting using a hybrid CNN-BiLSTM model · Engineering Applications of Artificial Intelligence · 2023 · 231 citations
• Short-term inter-urban traffic forecasts using neural networks · International Journal of Forecasting · 1997 · 216 citations
• A Bayesian network based framework for real-time crash prediction on the basic freeway segments of urban expressways · Accident Analysis & Prevention · 2011 · 208 citations

Most recent work

• GCMNet: A global context Mamba network for long-term time series forecasting · Pattern Recognition · 2026
• An integrated mutual-information clustering and deep neural network framework for forecasting travel behaviour under flexible working arrangements · Multimodal Transportation · 2026
• Towards fully automated city operations: Integrating agentic AI with urban digital twins · Computers Environment and Urban Systems · 2026
• Predicting Glare-Related Traffic Outcomes with Transformer-Based Explainable Tabular Deep Learning · Data Science for Transportation · 2026
• Towards Sustainable Urban Mobility: Interpretable Machine Learning for Bike-Sharing Inflow and Outflow Prediction · Data Science for Transportation · 2026
• The nonlinear impact of road safety policy implementation on the severity of road traffic crashes: A fusion of deep learning and Bayesian random parameter methods · Accident Analysis & Prevention · 2026
• Multimodal deep learning for tourism demand forecasting · Tourism Management · 2026
• Traffic speed prediction based on stacked denoising and lightweight Transformer · Discover Artificial Intelligence · 2026
• Deep heterogeneity learning for cross-city transit forecasting: a differentially private federated framework with mixture-of-experts and seasonal decomposition · Frontiers in Future Transportation · 2026
• Traffic flow prediction based on spatiotemporal feature fusion and Graph Convolutional Network · Engineering Computations · 2026

Related topics in Engineering