130 Big Data in Urban Traffic Management: Current Applications, Challenges, and Future Directions Jiaqi Qiu Qingdao University of Science and Technology, Qingdao, Shandong, China qiujiaqiqiqi@163.com Abstract. The integration of big data analytics represents a pivotal paradigm shift, moving urban traffic governance from traditional reactive models to proactive, intelligent systems. Through a comprehensive and systematic review of contemporary research literature, this paper synthesizes the current state of the field. It examines prominent applications, such as adaptive signal control and smart parking, and delineates their transformative benefits in reducing congestion, minimizing vehicle emissions, and enhancing overall urban mobility. The review then critically analyzes the persistent challenges that impede widespread implementation, primarily the foundational issues of data quality and integrity, the operational constraints of partial connectivity, and the critical privacy risks that can erode public trust. The significance of this research lies in its dual contribution: it informs the academic community by structuring existing knowledge and identifying crucial research gaps, while simultaneously providing actionable, data-driven insights for policymakers and urban planners. Conclusively, the paper outlines promising future directions, emphasizing the shift towards deep reinforcement learning for holistic network-wide optimization and the imperative of developing resilient, secure data ecosystems that are both efficient and trustworthy. Keywords: Big Data; Intelligent Transportation Systems; Traffic Congestion; Traffic Flow Prediction. 1. Introduction The rapid acceleration of urbanization has placed immense and unsustainable strain on urban traffic systems worldwide. This has led to a cascade of severe challenges, including chronic traffic congestion, increased environmental pollution from vehicle emissions, and a troubling rise in safety incidents [1]. Traditional traffic management strategies, often reliant on static infrastructure like fixed-time signal plans and sparse loop detectors, have proven increasingly inadequate. They typically operate in a reactive manner, addressing congestion only after it has formed, rather than anticipating and preventing it [2]. This reactive posture severely limits the potential for creating truly efficient, safe, and sustainable urban environments. In this context, the paradigm of Big Data analytics has emerged as a transformative force in modern urban traffic governance. By harnessing a multitude of data sources—from vehicle GPS trajectories [3], Internet of Things (IoT) sensor networks, and video surveillance to mobile payment records—Big Data enables a holistic and real-time understanding of complex urban mobility dynamics [4]. This technological shift facilitates a fundamental move from reactive problem-solving to proactive and predictive governance. The goal is no longer just to manage traffic, but to actively optimize network-wide efficiency, enhance public safety, and ultimately improve the daily quality of life for urban residents [2, 5]. As the field rapidly matures, a vast body of research has emerged. However, existing reviews often focus on specific sub- domains, such as the application of Big Data within public transit systems [6], or delve into singular analytical techniques like Graph Neural Networks for traffic forecasting [7]. A holistic overview that connects the diverse applications, their inherent challenges, and forward-looking research trajectories remains a critical need. This paper, therefore, provides a comprehensive and up-to-date review of the applications, advantages, challenges, and future directions of Big Data technology across the broader spectrum of urban traffic governance. To achieve this, the review first synthesizes key applications, from traffic flow prediction to advanced adaptive signal control [8] and smart parking [9]. It then critically examines significant implementation hurdles, including data quality, the complexities of partial connectivity [10], and overall system resilience [11]. By bridging these elements, this review 131 seeks to offer a valuable reference for researchers and practitioners dedicated to building the next generation of intelligent transportation systems. 2. Applications of Big Data in Urban Traffic Governance The effectiveness of big data applications fundamentally relies on the ability to capture and process vast and detailed traffic information. As reviewed by Lu et al. [6], this is achieved through various data collection technologies, where key sources include Automated Fare Collection (AFC) systems, vehicle Global Position System (GPS) trackers, and data from smartphones. These sources provide detailed insights into passenger travel and vehicle operations, forming the foundation for the diverse applications discussed below. 2.1 Traffic Flow Prediction Accurate and timely traffic flow prediction is a cornerstone of intelligent transportation systems. However, as noted by Salazar-Castañeda et al. [1], this task is inherently challenging due to the complex, non-linear, and stochastic nature of traffic, as well as the intricate spatio-temporal dependencies within road networks. Machine learning and deep learning methods have proven highly effective in tackling these complexities by training models on historical traffic data. Common methods include Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. For instance, Wu et al. proposed a hybrid model based on CNN and GRU that effectively captures spatial and temporal features of traffic flow data for high-precision predictions. The systematic survey [7] further highlights the evolution of these predictive models, noting the application of advanced architectures such as the Graph Convolutional Recurrent Neural Network (GCRN) to better capture the complex spatio-temporal dependencies inherent in traffic networks. These advanced models are crucial because, as stated in a comprehensive survey by Zhu et al. [2], fast and precise analysis of massive traffic data allows traffic management departments to predict and understand traffic flow in real-time, forming a basis for subsequent control actions. Case studies based on different city datasets demonstrate the effectiveness of big data-driven traffic flow prediction systems[12]. 2.2 Traffic Signal Control Optimization Real-time traffic data is used to adjust traffic light durations, thereby reducing congestion. Deep reinforcement learning models like DQN are applied to optimize traffic signal control, with studies like Shang et al. [12] demonstrating through simulations their effectiveness in reducing congestion. Beyond single-intersection optimization, a significant frontier is enhancing cooperation between multiple traffic signals. Li et al. [8] propose a novel multi-agent reinforcement learning (MARL) method where each traffic signal acts as an agent. By introducing a knowledge-sharing communication protocol, each agent can access a collective understanding of the entire traffic environment, leading to more coordinated, network-wide optimal control. This approach, leveraging multi-agent deep reinforcement learning, has demonstrated significant efficiency in controlling large- scale transportation networks. Studies have shown its direct impact on real-world traffic improvement, such as reducing average vehicle travel time and shortening queue lengths at intersections [8]. Furthermore, by creating smoother traffic flows, these advanced control strategies contribute significantly to environmental benefits, including a notable reduction in vehicle emissions and overall energy consumption [12]. However, a significant real-world challenge remains the partially connected environment. Research by Jia et al. [10] on adaptive signal control specifically addresses this issue, highlighting the critical need to accurately estimate traffic states, such as queue lengths, from a limited number of connected probe vehicles to make effective control decisions. 132 2.3 Traffic Congestion Management Big data analytics are employed to proactively identify congestion-prone areas and bottlenecks. By analyzing data on traffic patterns and vehicle movements, systems can predict future traffic patterns and issue timely warnings to drivers and authorities [11]. Neilson et al. analyzed traffic flow data to identify congestion in real-time and used mobile applications to alert drivers [4]. Big data- driven traffic diversion strategies are also implemented in intelligent transportation systems, where intelligent routing algorithms automatically adjust vehicle paths based on real-time traffic flow to effectively alleviate congestion [13]. Beyond identifying aggregate congestion, big data allows for a granular analysis of passenger behavior to understand its root causes. For example, Lu et al. [6] discuss the identification of extreme travelers, such as passengers who travel significantly earlier than average (the early birds) or those who take excessively long trips. Understanding the patterns of these specific user groups enables transit authorities to better manage recurring, predictable congestion. According to Wang et al. [3], trajectory-based data allows for the identification of congestion hotspots through stop-point density analysis, providing a more granular understanding of congestion patterns. This method supports the development of dynamic diversion strategies and improves the responsiveness of congestion management systems. 2.4 Traffic Safety Enhancement Big data analytics play a multi-faceted role in enhancing traffic safety, moving beyond reactive responses to proactive prevention. A primary application is the identification of high-risk areas. By analyzing diverse data sources, systems can pinpoint accident hotspots and enable authorities to implement targeted interventions, such as adjusting speed limits or improving road design[11]. Furthermore, big data enables the prediction and prevention of accidents. As discussed by Zhu et al. [2], advanced analytics on real-time and historical data can effectively predict the occurrence of traffic accidents. This proactive capability is complemented by analysis of vehicle trajectory data, which can detect abnormal driving behaviors (e.g., erratic movements, sudden stops) that often precede collisions [3]. In the event of an accident, this real-time capability transforms emergency response. Instead of relying solely on witness reports, big data systems can automatically transmit precise collision details—such as GPS coordinates from in-vehicle systems [13] and potential impact severity—directly to public security agencies. Simultaneously, the system analyzes real-time traffic data to calculate the optimal, congestion-free route for emergency vehicles, a core component of enhancing overall transportation infrastructure resilience [11]. This data-driven approach dramatically reduces dispatch and travel times, thus greatly improving emergency rescue efficiency [5]. 2.5 Smart Parking Management A significant contributor to urban congestion and driver frustration is the search for parking. Smart parking has emerged as a key application of big data and AI in urban governance. As detailed by Kumar et al. [9], AI-powered smart parking systems utilize IoT sensors and live video streams to provide real-time information on parking spot occupancy. Machine learning models, such as Random Forest and Recurrent Neural Networks, are then used to predict future parking availability. This allows for automated spot allocation, personalized recommendations for drivers via mobile apps, and seamless integration with other urban mobility systems, ultimately reducing the time spent circling for parking and improving overall traffic flow. 3. Challenges of Big Data in Urban Traffic Governance While the potential of big data in revolutionizing urban traffic governance is undeniable, its implementation faces significant and interconnected challenges that can impede its effectiveness and widespread adoption. These challenges span from the fundamental quality of data to the complexities of real-world operational environments and critical ethical considerations. The axiom garbage in, 133 garbage out is particularly pertinent in big data analytics. The reliability of advanced models for traffic prediction [1] and control [8] is fundamentally dependent on the quality of the input data. However, real-world data streams are often plagued by issues such as missing values, inaccuracies from sensor malfunctions, and inconsistencies across different data sources [2]. This poor data quality can lead to flawed analysis and unreliable predictions, directly undermining the trustworthiness of the system and potentially leading to suboptimal or even counterproductive traffic management decisions. Consequently, a substantial portion of resources must be dedicated to data preprocessing and cleaning, which remains a complex and time-consuming prerequisite for meaningful analysis. A major operational hurdle is the partially connected environment, where only a fraction of vehicles are connected vehicles (CVs) capable of transmitting data. This incomplete penetration rate means that the available data provides only a fragmented snapshot of the overall traffic state. For applications like adaptive signal control, accurately estimating queue lengths and arrival rates becomes critically difficult with sparse data [10]. This uncertainty directly impacts the performance of optimization algorithms, as decisions made based on an incomplete picture may not be globally optimal. As a result, the full potential of advanced control strategies cannot be realized until a higher penetration of connected technologies is achieved, limiting their current real-world impact. The collection of vast amounts of granular data, such as individual vehicle trajectories [3] and mobile payment information, raises significant privacy and security concerns. Such data, if improperly handled, can reveal sensitive personal patterns and expose individuals to risks of surveillance or misuse. This potential for privacy infringement can erode public trust, leading to resistance against data-sharing initiatives crucial for the system’s success [4]. Furthermore, centralized data systems present an attractive target for cyber-attacks, with the risk of data breaches or malicious manipulation of traffic control systems posing a direct threat to public safety. Therefore, the implementation of these technologies is severely constrained by the need to develop robust privacy-preserving frameworks and secure data governance protocols. In summary, the path to fully leveraging big data in urban traffic is obstructed by a triad of fundamental challenges. Data quality and integrity issues form the foundational barrier, while partial connectivity presents a significant operational constraint. Layered on top are the critical privacy and security concerns that dictate the social and ethical boundaries of these systems. Overcoming these obstacles is not merely a technical task; it requires a concerted effort that combines technological innovation with robust institutional policies and regulatory frameworks. Addressing these challenges systematically is essential for building a future where intelligent transportation systems are not only efficient but also reliable, secure, and trustworthy, setting the stage for the future research directions discussed next. 4. Future Directions Building upon the identified applications and challenges, the future of big data in urban traffic governance will focus on developing more integrated, robust, and trustworthy systems. A key direction is the fusion of multi-modal data, integrating disparate sources like vehicle GPS, public transit records, and real-time weather into unified analytical frameworks. Concurrently, research must address the partial connectivity challenge by creating models robust to sparse data, potentially leveraging generative techniques to supplement incomplete information. To ensure real-time performance and scalability across entire cities, an architectural shift towards Edge Computing is anticipated, moving data processing closer to the source. Critically, this evolution must be underpinned by a foundation of trust. This involves integrating privacy-preserving techniques, such as Federated Learning and Differential Privacy, to safeguard citizen data. Furthermore, the adoption of Explainable AI (XAI) will be essential to make algorithmic decisions transparent and understandable to city planners, fostering confidence and facilitating the adoption of next-generation traffic management systems. 134 5. Conclusion This review has comprehensively examined the application of big data technology as a transformative force in advancing sustainable urban development. By enabling a paradigm of proactive and intelligent traffic management, it moves beyond the limitations of traditional, reactive methods. As demonstrated, its applications directly contribute to sustainability goals, from optimizing network-wide signal control to simultaneously reducing congestion and vehicle emissions, to enhancing travel efficiency and public safety through predictive analytics and abnormal behavior detection. These advancements are foundational to creating the efficient, green, and resilient urban transportation systems of the future. However, realizing this full potential is contingent upon overcoming a triad of significant challenges: the foundational barrier of data quality, the operational constraint of partial connectivity, and the critical ethical boundaries of privacy and security. By systematically identifying and analyzing these hurdles, this review provides valuable insights for both practitioners and researchers. For city managers and policymakers, it offers a data-driven reference for strategic planning and risk-aware implementation. For the academic community, it identifies key avenues for future investigation, underscoring the urgent need for innovation in areas such as advanced data fusion, privacy-preserving machine learning, and resilient communication protocols. In conclusion, while big data has already unlocked substantial improvements in urban mobility, its journey is one of continuous evolution. Addressing the technical and ethical hurdles identified in this review is not a solitary task but requires a synergistic collaboration between researchers, policymakers, and industry stakeholders. Only through such a concerted effort can the promise of big data be fully realized, leading to the next generation of intelligent transportation systems that are not only efficient but also reliable, secure, and fundamentally trustworthy, thereby making a lasting contribution to the quality and sustainability of urban life. References [1] Medina-Salgado, B., Sánchez-DelaCruz, E., Pozos-Parra, P., & Sierra, J. E. (2022). Urban traffic flow prediction techniques: A review. Sustainable Computing: Informatics and Systems, 35, 100739. [2] Zhu, L., Yu, F. R., Wang, Y., Ning, B., & Tang, T. (2018). Big data analytics in intelligent transportation systems: A survey. IEEE transactions on intelligent transportation systems, 20(1), 383-398. [3] Wang, Y., Liu, Y., & Yang, X. (2025). A Novel Method for Traffic Parameter Extraction and Analysis Based on Vehicle Trajectory Data for Signal Control Optimization. Applied Sciences, 15(13), 7155. [4] Neilson, A., Daniel, B., & Tjandra, S. (2019). Systematic review of the literature on big data in the transportation domain: Concepts and applications. Big Data Research, 17, 35-44. [5] Liu, Y. (2018, January). Big data technology and its analysis of application in urban intelligent transportation system. In 2018 International Conference on Intelligent Transportation, Big Data & Smart City (ICITBS) (pp. 17-19). IEEE. [6] Lu, K., Liu, J., Zhou, X., & Han, B. (2020). A review of big data applications in urban transit systems. IEEE Transactions on Intelligent Transportation Systems, 22(5), 2535-2552. [7] Jiang, W., & Luo, J. (2022). Graph neural network for traffic forecasting: A survey. Expert systems with applications, 207, 117921. [8] Li, Z., Yu, H., Zhang, G., Dong, S., & Xu, C. Z. (2021). Network-wide traffic signal control optimization using a multi-agent deep reinforcement learning. Transportation Research Part C: Emerging Technologies, 125, 103059. [9] Shalini, M. K., Hanumanthappa, J., KS, S. K., & SP, S. P. (2025). AI-Powered Hybrid Smart Parking: Optimizing Parking Management Across Diverse Applications in Smart Cities. Procedia Computer Science, 258, 1524-1535. [10] Shaocheng, J. I. A., Wong, S. C., & Wai, W. O. N. G. (2025). Adaptive signal control at partially connected intersections: A stochastic optimization model for uncertain vehicle arrival rates. Transportation Research Part B: Methodological, 193, 103161. 135 [11] Foltin, P. D. P., Nagy, P. D. J., & Ondryhal, P. D. V. (2023). Using Big Data Analysis in increasing transportation infrastructure resilience. Transportation Research Procedia, 74, 1508-1515. [12] Shang, W. L., Song, X., Xiang, Q., Chen, H., Elhajj, M., Bi, H., … & Ochieng, W. (2025). The impact of deep reinforcement learning-based traffic signal control on Emission reduction in urban Road networks empowered by cooperative vehicle-infrastructure systems. Applied Energy, 390, 125884. [13] Gaol, F. L., Alam, P., Franklyn, M. B., Angke, K., & Matsuo, T. (2023). Traffic Light System With Embedded GPS (Global Positioning System) and GSM (Global System for Mobile Communications) Shield. International Journal of Ambient Computing and Intelligence (IJACI), 14(1), 1-13.