Urban Traffic State Sensing and Analysis Based on ETC Data: A Survey

Abstract

Urban traffic management faces challenges, including inadequate sensing capabilities and insufficient operational status evaluation. The rapid expansion of electronic toll collection (ETC) systems from highways to urban roads provides new opportunities to address these issues. The vast amount of “dormant” ETC data contains rich traffic information that urgently needs to be deeply mined and effectively utilized. This paper reviews the research status, key technologies, and development trends of urban traffic state sensing and analysis technologies based on ETC data. In terms of technological development, ETC systems have evolved from simple toll collection tools to comprehensive traffic management platforms, featuring unique advantages such as accurate vehicle identification, extensive spatiotemporal coverage, and stable data quality. ETC data-based traffic sensing technologies encompass traffic state representation at microscopic, mesoscopic, and macroscopic levels, enabling comprehensive sensing from individual vehicle behavior to overall network operations. The construction of multi-source data fusion frameworks enables effective complementarity between ETC data, floating car data, and video detection data, significantly improving traffic state estimation accuracy. In practical applications, ETC data has demonstrated enormous potential in real-time monitoring and signal control optimization, traffic prediction and artificial intelligence technologies, environmental impact assessment, and other fields. Meanwhile, ETC data-based urban traffic management is transitioning from passive responses to proactive prediction, from single functions to comprehensive services, and from isolated systems to integrated platforms. Looking toward the future, the deep integration of emerging technologies, such as vehicle–road networking, edge computing, and artificial intelligence, with ETC systems will further promote the intelligent, refined, and precise development of urban traffic management.

1. Introduction

1.1. Research Background

China has entered a new era of strategic development characterized by building a strong transportation system, rapid urbanization, economic and social transformation, intelligent vehicle–road cooperative control, and digital infrastructure development for transportation facilities. Against this historical backdrop, traffic management technologies face unprecedented challenges and opportunities. In August 2020, the China Association for Science and Technology released ten scientific questions with guiding significance for scientific development, with the sixth frontier scientific question explicitly asking: “How can digital transportation infrastructure promote the development of autonomous driving and vehicle-road cooperation?” ETC-based traffic sensing and analysis methods represent one of the important research directions established to address this question [1].

With the continuous acceleration of urbanization, urban transportation worldwide is facing unprecedented challenges. Under such high-intensity traffic demand pressure, traditional traffic management and sensing methods have exposed numerous limitations. Existing traffic detection technologies mainly rely on loop detectors, video surveillance, microwave radar, and other equipment. Although these technologies perform well in specific scenarios, they generally suffer from limited coverage, high equipment maintenance costs, data quality susceptible to environmental influences, and difficulty in achieving large-scale continuous monitoring. Particularly in complex urban traffic environments, traditional sensing methods often fail to provide sufficiently accurate and timely traffic state information, affecting the scientific validity and effectiveness of traffic management decisions.

In this context, the rapid development of electronic toll collection (ETC) systems has brought important opportunities for solving urban traffic sensing challenges. ETC systems, with their unique technological advantages, can achieve precise vehicle identification and data collection while vehicles pass at high speeds, featuring characteristics such as high identification accuracy, extensive spatiotemporal coverage, stable data quality, and relatively low operational costs.

1.2. Research Opportunities

Although ETC systems have achieved large-scale deployment and application nationwide, their primary application has been in highway scenarios, with their potential value in urban traffic management far from being fully exploited. Current ETC applications mainly focus on toll collection functions, while the massive high-quality traffic data resources generated remain largely “dormant,” lacking deep mining and systematic utilization.

From the perspective of technological development, ETC system construction has provided important infrastructure support for urban traffic digitization transformation. As of 2023, China has built and networked 29,500 ETC gantries and 250,000 license plate recognition devices on highways, successfully constructing and optimizing the world’s largest highway ETC system that operates safely and stably, serving nearly 240 million ETC users. Meanwhile, China has approximately 12,000 toll stations with 96,436 toll lanes, including 42,554 ETC lanes. This massive infrastructure network provides a solid hardware foundation for data collection and analysis.

From a policy perspective, the government has given high importance and strong support to the application of ETC technology in urban transportation. In February 2021, the Ministry of Transport officially issued the “Notice on Launching Pilot Work for ETC Smart Parking City Construction,” clearly proposing to strengthen the integrated development of “ETC + Internet” industries through the deep integration of emerging technologies such as big data, artificial intelligence, and 5G with ETC technology. The policy document particularly emphasizes creating a comprehensive ETC + industrial chain, including ETC + IoT (Internet of Things) sensing, ETC + intelligent connected communication, ETC + big data platforms, ETC + static traffic management, and ETC + vehicle owner services, forming a new data-driven urban traffic management service model [2].

From a market demand perspective, urban traffic management departments urgently need more precise, real-time, and comprehensive traffic state sensing methods. Traditional traffic survey methods, such as manual observation and questionnaire surveys, are not only costly and inefficient but also have difficulty meeting the requirements of refined and intelligent modern urban traffic management. ETC data-based traffic sensing technology can achieve large-scale, high-precision traffic state monitoring at a relatively low cost, providing scientific and reliable data support for traffic management decisions.

From a technological maturity perspective, ETC systems have achieved high maturity in hardware equipment, communication protocols, and data processing through years of development and improvement. Particularly, the standardized application of Dedicated Short-Range Communication (DSRC) technology has provided technical assurance for reliable ETC data collection and transmission. Meanwhile, the rapid development of emerging technologies, such as big data processing, artificial intelligence, and cloud computing, has created favorable conditions for the deep analysis and intelligent applications of ETC data [3,4].

1.3. Research Problems and Contributions

Currently, although research on ETC data-based traffic sensing and analysis technologies has made certain progress, it still faces numerous challenges. In theoretical research, traffic flow theories and analysis methods targeting ETC data characteristics are not yet sufficiently developed, lacking systematic theoretical frameworks. In technical applications, most research remains at the level of data processing and simple statistical analysis, lacking deep, intelligent analysis and prediction capabilities. In practical applications, although some cities have conducted related pilot work, the application scope is relatively limited, and the promotion and replication of successful experiences need to be strengthened.

Based on the above background analysis, this paper aims to review the research status, key technologies, and development trends of ETC data-based urban traffic state sensing and analysis technologies, focusing on examining and analyzing the following three core research areas.

First, how existing research constructs multi-level traffic sensing methodology systems based on ETC. This includes reviewing the complete feature extraction research frameworks covering three levels: cross-section traffic state representation, road segment traffic state representation, and road network traffic state representation, ranging from microscopic vehicle individual behavior recognition to mesoscopic road segment traffic state analysis and macroscopic overall road network operation assessment.

Second, how current studies achieve the effective fusion of ETC data with other traffic data sources. This review focuses on comprehensive traffic analysis methods based on ETC+GPS data, including deep feature extraction technologies for vehicle operation state monitoring, travel time distribution analysis, and road segment operation reliability assessment.

Third, how researchers construct urban traffic dynamic balance diagnosis systems supported by ETC and big data. By examining the deep fusion and intelligent analysis of multi-source data, this review analyzes the transformation from traditional passive-response mode to proactive-prediction mode, and the establishment of information-led, intelligence-driven multi-system collaborative control mechanisms.

1.4. Literature Search Methodology

This comprehensive narrative review employed a multi-database literature search to identify the relevant studies on ETC data-based urban traffic state sensing and analysis technologies. The primary databases searched included Web of Science Core Collection, IEEE Xplore Digital Library, ScienceDirect, Transportation Research International Documentation (TRID), and Engineering Index (EI Compendex) to ensure the comprehensive coverage of the international literature in transportation engineering, computer science, and intelligent systems. Given the significant developments in ETC technology in China and other regions, supplementary searches were conducted in regional databases, including China National Knowledge Infrastructure (CNKI) and relevant conference proceedings from major venues such as the IEEE Intelligent Transportation Systems Conference, Transportation Research Board Annual Meeting, and International Conference on Intelligent Transportation Systems.

The search strategy utilized comprehensive combinations of key terms across multiple domains: ETC-related terms (“Electronic Toll Collection,” “ETC system,” “DSRC,” “On-Board Unit,” “Road-Side Unit”); traffic sensing terms (“traffic state sensing,” “traffic flow detection,” “speed estimation,” “congestion detection,” “traffic monitoring”); data analysis terms (“data fusion,” “feature extraction,” “spatiotemporal analysis,” “machine learning,” “artificial intelligence,” “graph neural networks”); urban applications (“urban traffic management,” “intelligent transportation systems,” “smart transportation,” “traffic prediction,” “signal control optimization”); and integration concepts (“multi-source data,” “sensor fusion,” “big data analytics,” “edge computing,” “vehicle-road cooperation”). Boolean operators (AND, OR) were employed to create targeted search strings, with language filters applied to include studies in English and Chinese.

The temporal scope primarily focused on publications from 2010 to 2025 to capture the evolution from basic ETC toll collection systems to advanced urban traffic management applications, while foundational earlier works (1998-2009) were selectively included when historically significant for understanding ETC technology development trajectory.

The literature selection process involved multiple stages: (1) initial automated screening based on title relevance and abstract content; (2) full-text review for studies meeting inclusion criteria of ETC technology application in urban traffic contexts; (3) quality assessment focusing on peer-reviewed publications, conference papers from recognized venues, and technical reports from established research institutions; (4) backward and forward citation tracking to identify additional relevant studies. Inclusion criteria specifically targeted studies addressing ETC data processing, urban traffic state analysis, multi-source data fusion involving ETC, artificial intelligence applications in ETC systems, and practical implementations in urban environments. Studies focusing solely on highway applications without urban relevance, purely theoretical models without practical validation, and non-English/Chinese publications were excluded.

The process emphasized thematic organization around key technical and application domains: (1) ETC system architecture and technology principles; (2) data processing and quality control methodologies; (3) traffic state feature extraction across microscopic, mesoscopic, and macroscopic levels; (4) multi-source data fusion frameworks and algorithms; (5) artificial intelligence and machine learning applications; (6) urban traffic management implementation cases; (7) emerging technologies and future integration directions. This organizational structure facilitates an understanding of both technological foundations and practical applications while identifying research gaps and future opportunities in ETC-based urban traffic management.

1.5. Paper Structure

This paper adopts a logical structure progressing from theory to practice and from current status to prospects, comprising seven sections: Section 1 introduces the research background and core problems; Section 2 reviews the development history and application status of ETC systems; Section 3 constructs a multi-level traffic sensing and feature extraction framework based on ETC data; Section 4 explores traffic state analysis and assessment methods in depth; Section 5 demonstrates the practical value of ETC data in urban traffic management through rich application cases; Section 6 analyzes the current technical challenges and identifies future research directions; Section 7 summarizes the main contributions and provides prospects for future development.

2. ETC Systems: Historical Development and Current Applications

2.1. ETC Technology Principles and System Architecture

Electronic toll collection (ETC) represents an intelligent transportation subsystem that leverages Dedicated Short-Range Communication (DSRC) technology to automate toll collection processes. The system enables wireless data communication between vehicles and toll stations, facilitating automatic vehicle identification and toll data exchange. Through computer network processing, ETC achieves fully automated, non-stop toll collection capabilities.

The ETC system primarily consists of three components: (1) On-Board Unit (OBU): an electronic tag installed on the vehicle’s windshield that stores vehicle information and user account data; (2) Road-Side Unit (RSU): antenna equipment installed at toll stations or roadside locations responsible for communicating with OBUs; (3) Data Center: responsible for data processing, billing settlement, and system management. From the perspective of technological architecture evolution, early ETC systems were primarily based on a single DSRC technology. With technological advancement, modern ETC systems have evolved into comprehensive platforms integrating multiple technologies. This evolution of system architecture not only enhances communication reliability but also provides a technical foundation for subsequent multi-source data fusion and intelligent analysis [5].

2.2. Global ETC Development History Overview

The development of ETC technology worldwide exhibits distinct gradual characteristics, experiencing an extended development cycle from technical validation to large-scale application while accumulating rich experiences and mature technical solutions [6,7,8].

As a pioneer in ETC technology, the United States’ development experience offers significant reference value. American highways are generally toll-free, with toll roads accounting for less than one-tenth of the total mileage. However, ETC technology has been fully applied and validated on the limited toll segments. A key characteristic of the American ETC system is regional development, with states initially building their own systems and later achieving regional integration through negotiations. E-ZPass became the unified brand for 16 northeastern states, while Sun-Pass covers three southern states: Florida, Georgia, and South Carolina. The success of American ETC development lies in its flexible business models and user-friendly design concepts. Levinson and Chang found in their research that appropriate ETC lane configuration and reasonable fee discounts are key factors in improving overall system efficiency [9]. American experience demonstrates that ETC system success depends not only on technology itself but more importantly on establishing reasonable business models and incentive mechanisms.

Japan’s ETC development trajectory demonstrates a complete path from technology introduction to comprehensive adoption, providing valuable insights for understanding ETC technology promotion patterns. In 1997, Japan’s ETC toll collection system conducted its first trial operation on the Odawara-Atsugi Road. Through long-term technical accumulation and market cultivation, it eventually achieved large-scale adoption. Research by Ohya et al. indicates that mixed lane configurations and dynamic lane switching strategies can maximize ETC system benefits [10]. The distinctive features of Japan’s ETC development include continuous technological upgrades and systematic policy support. The national promotion of the ETC toll collection system began in late November 2000, but the initial adoption was relatively slow due to complex credit card application procedures and high OBU equipment prices. By implementing equipment subsidies, fee discounts, convenient services, and other measures, Japan upgraded to ETC2.0 in 2015, achieving more intelligent management through big data and internet technologies. By 2017, the number of vehicles equipped with ETC exceeded 70 million, with ETC usage rates surpassing 90%.

Singapore’s Electronic Road Pricing (ERP) system represents an innovative application of ETC technology in urban traffic management. Officially implemented in September 1998, this system employs dynamic pricing strategies, establishing different toll standards based on road traffic speeds and regulating traffic demand through economic leverage. Singapore has successfully applied ETC technology to urban intelligent transportation systems, with ETC parking and ETC payments becoming routine aspects of Singaporean travel. The success of this model validates the enormous potential of ETC technology in urban traffic demand management.

China’s ETC development exhibits distinct policy-driven characteristics, demonstrating the advantages of centralized coordination and extraordinary development capabilities. Unlike the gradual development abroad, China’s ETC construction has followed a leapfrog development path. China’s ETC development can be divided into three phases: the pilot exploration period (1998–2010), the steady development period (2010–2019), and the rapid adoption period (2019–present). In 1998, the ETC toll collection system began its trial implementation in Guangdong Province. After more than 20 years of development, it gradually expanded from regional pilots to nationwide networking. This development process reflects the gradual and systematic characteristics of China’s new technology promotion and application. Another important feature of China’s ETC development is the uniformity of the technical standards and system compatibility. Unlike the American model of states building independent systems before integration, China has emphasized the formulation and implementation of unified national standards from the beginning. The advantages of this unified standard approach were fully demonstrated during the nationwide networking process, providing a technical foundation for data interconnection and cross-regional applications [11].

2.3. Current Application Status of ETC in Urban Transportation

With the successful application of ETC systems on highways, their application in urban transportation has gradually expanded, showing diversified and integrated development trends. The complexity of urban transportation environments poses new challenges for ETC technology while providing a greater scope for functional expansion.

Hu Xiaojuan pointed out in her research that ETC systems play important roles in improving traffic efficiency, promoting multi-departmental coordination, and innovating service models in intelligent transportation [12]. This perspective accurately summarizes the core values of ETC technology in urban transportation applications. Unlike the relatively simple application scenarios on highways, the diversity of urban transportation requires ETC systems to possess stronger adaptability and scalability.

The deep application of ETC + parking scenarios represents an important direction for urban transportation intelligence. Parking facilities at airports, hospitals, shopping centers, and other locations have introduced ETC contactless payment, achieving non-stop rapid passage and significantly improving user experience and operational efficiency. Zheng Jianhu et al. designed an intelligent parking management system based on ETC technology, achieving unmanned parking management through system integration and significantly improving parking management efficiency and user satisfaction [13]. The success of this application model demonstrates that ETC technology has enormous application potential in urban static traffic management.

The integrated application of ETC + transit systems embodies the technical implementation of public transportation priority development strategies. Applying ETC technology in bus lane management and transit signal priority enables precise vehicle identification and priority control, improving transit operational efficiency. Some cities have begun piloting ETC applications in transit vehicle management, achieving transit vehicle priority through coordination with signal control systems. The parallel vehicle networking concept proposed by Wang Xiao et al. provides a theoretical framework for the deep integration between ETC and transit systems [14].

The intelligent management of ETC + logistics and delivery demonstrates the application value of ETC technology in freight transportation management. Utilizing ETC technology for truck passage management, route monitoring, and transportation optimization not only improves logistics efficiency but also provides effective means for urban freight management. Yin Zhijie et al. studied the innovative applications of ETC in multi-scenario applications, particularly emphasizing the important role of ETC technology in addressing pain points such as cumbersome payments and inefficient management in urban transportation, parking, and refueling processes [15].

The emergence of multi-source data fusion trends marks the evolution of ETC applications toward higher levels. Increasingly more projects adopt “ETC+video+radar+MEC” multi-source fusion sensing solutions. Compared to single video surveillance or microwave radars, multi-source fusion provides higher detection accuracy and anti-interference capabilities. Fusion sensing solutions accomplish vehicle identity identification and positioning, multi-target collaborative tracking, and traffic anomaly event determination through data cleaning, coordinate transformation, feature extraction, and structured fusion based on toll data and multi-source sensing. This trend of technological fusion indicates that ETC is evolving from a single identification technology to a comprehensive sensing platform.

Table 1. Comparison of application effects of different sensing technologies.

Performance Dimension	Video Camera Technology	Millimeter Wave Radar Technology	ETC Multi-Sensor Fusion Technology
System Architecture	Network connection to host computer or cloud	Network connection to host computer or cloud	Cloud-based architecture with edge computing support
Detection Coverage	Generally covers 3–4 lanes	Up to 12 bidirectional lanes	Unlimited lane coverage capability
Core Detection Capabilities	Vehicle identification Target characteristics analysis Visual evidence collection	Target characteristics’ detection Precise position coordinates Motion parameter measurement	Vehicle identity tracking Position coordinate detection Travel path reconstruction Multi-modal data integration
Environmental Adaptability	Performance varies with lighting and weather conditions	High detection accuracy, independent of environmental conditions	High detection accuracy, independent of environmental conditions
Data Output Types	Trajectory data (position, speed, vehicle type) Traffic flow parameters (volume, speed, occupancy, queue length) Traffic event detection (parking, speeding, lane changing, wrong-way driving, congestion)	Motion parameters and target classification Distance and position measurements Basic traffic flow statistics	Comprehensive traffic parameters Vehicle identity and ownership information Historical transaction records Cross-regional travel patterns Small-target tracking (pedestrians) Enhanced event detection capabilities
Data Processing Architecture	Data transmission to cloud	Data transmission to cloud	Supports both local processing and cloud computing
Computing Capabilities	Limited computing power and non-upgradeable hardware	Limited computing power and non-upgradeable hardware	Scalable edge computing with video analysis and sensor fusion capabilities
Algorithm Flexibility	Fixed, single-purpose algorithms	Fixed, single-purpose algorithms	Flexible multi-algorithm configuration with site-specific training capabilities and continuous optimization support
Speed Measurement Accuracy	Large measurement errors due to visual estimation	High-precision radar-based measurement with small errors	High-precision radar-based measurement with small errors
Distance Measurement Performance	Poor accuracy for distance estimation	Good distance measurement capability	Excellent distance measurement with multi-sensor validation
Target Detection Reliability	Environmental sensitivity with higher miss detection rates	Dual-method detection with improved reliability	Triple-method detection, achieving the highest detection rates
Vehicle–Infrastructure Communication	No communication capability	No communication capability	Bidirectional communication, enabling safety warnings and traffic coordination
Performance Upgrade Space	Hardware-limited and non-upgradeable	Hardware-limited and non-upgradeable	Software-based upgrades with continuous algorithm optimization

summarizes the comparison of different sensing technology application effects.

However, ETC applications in urban transportation still face several challenges. First is the coverage issue: although highway ETC adoption rates are high, coverage on urban roads remains limited. Second is the standardization issue: ETC applications in different cities and scenarios lack unified standards, affecting system interconnection and interoperability. Finally, data security and privacy protection issues have become important challenges as application scenarios expand, requiring a balance between data utilization and privacy protection.

From a development perspective, ETC applications in urban transportation exhibit several important characteristics: First, the diversification of application scenarios, expanding from single toll collection functions to multiple domains including parking, transit, and logistics; second, deepening technological integration, with an increasingly deep fusion between ETC and technologies such as video, radar, and communications; third, intelligent service models, with intelligent services based on big data and artificial intelligence becoming the development direction; fourth, refined management approaches, with ETC data providing data support for refined traffic management.

These development trends indicate that ETC technology is transforming from traditional toll collection tools to the important components of urban intelligent transportation systems, providing crucial technical support for constructing smart city transportation systems. As technology continues to mature and applications continue to expand, the role of ETC in urban traffic management will become more prominent, and its data value will be more fully explored and utilized.

3. ETC Data-Based Traffic Sensing and Feature Extraction

Building upon the technological foundations and application contexts established in the previous section, this section examines the core technical methodologies for extracting meaningful traffic information from ETC data.

Traffic sensing technology based on ETC data represents a significant technological breakthrough in the field of traffic information collection, with its core value lying in transforming traditional toll data into rich traffic state information. This transformation process not only demonstrates the innovative applications of data mining technology but more importantly provides new sensing methods for urban traffic management. Compared with traditional point detectors, ETC data possesses unique advantages, such as trackable vehicle identity, continuous spatiotemporal coverage, and stable data quality, providing a solid foundation for constructing comprehensive and multi-level traffic sensing systems.

3.1. ETC Data Structure and Characteristic Analysis

3.1.1. Data Types and Structures

Data generated by ETC systems are characterized by high structuralization and rich information content. These data not only record vehicle passage information but also contain deep-level traffic operation patterns and user travel characteristics.

The deep value mining of user card data has become an important direction for ETC data applications. User card data contains multi-dimensional information, including basic cardholder information (name, identification number, contact information), vehicle information (license plate number, vehicle type, color), and account information (balance and credit limit). These data provide the foundation for vehicle identity recognition and user behavior analysis while creating conditions for personalized traffic services and precise traffic management. From the perspective of data structure evolution, user data in modern ETC systems has evolved from simple identity identification to a foundation for user profiles containing diverse information such as travel preferences, consumption habits, and credit status.

The spatiotemporal characteristic analysis of transaction record data reflects the core value of ETC data in traffic sensing. Transaction records detail information for each passage, including entry time, exit time, entry number, exit number, travel path, toll amount, and other key elements. Weng Jiancheng et al. developed systematic information extraction technologies based on ETC electronic toll data, successfully achieving road segment travel speed calculation and traffic flow distribution analysis [16]. The significance of this research lies in validating the reliability and effectiveness of ETC transaction data in traffic parameter estimation.

Vehicle trajectory data reconstruction and application represent advanced forms of ETC data applications. By linking identification records from multiple ETC gantries, vehicle travel trajectories can be accurately reconstructed, providing valuable data resources for traffic flow analysis, route choice research, and travel behavior analysis. This not only demonstrates the technical feasibility of ETC data in large-scale traffic monitoring but more importantly showcases the enormous potential of trajectory data in discovering urban traffic patterns.

From the perspective of data structure development trends, ETC data is undergoing transformation from single-dimensional to multi-dimensional, from static recording to dynamic sensing, and from passive collection to active interaction. These transformations not only enrich the content and value of data but also provide a greater scope for the innovative development of traffic sensing technologies.

3.1.2. Data Quality Control

Although ETC data possesses significant advantages, such as high precision and strong real-time characteristics, comprehensive data quality control systems must still be established in practical applications to ensure data accuracy, completeness, and consistency. Data quality control is not only a technical issue but also a critical component affecting the reliability of the entire traffic sensing system.

The systematic development of data cleaning and preprocessing technologies reflects the maturity of ETC data applications. Data quality can be significantly improved through technical means, such as removing duplicate records, correcting temporal anomalies, and processing missing values. The establishment of anomaly data identification and processing mechanisms is crucial for ensuring ETC data reliability. By setting reasonable business rules and statistical thresholds, problematic data such as speed anomalies, time interval anomalies, and path anomalies can be effectively identified.

The systematic development of data cleaning and preprocessing technologies reflects the maturity of ETC data applications. Data quality can be significantly improved through technical means such as removing duplicate records, correcting temporal anomalies, and processing missing values. Xue Bingbing et al. proposed systematic quality control methods based on ETC toll data for real-time expressway traffic data imputation and state prediction, employing low-rank tensor decomposition theory to repair missing data [17]. Their approach utilized confusion matrices to determine the thresholds for identifying missing data and introduced three-dimensional tensor models with low-rank tensor completion methods for missing data repair. Compared with traditional Lagrangian interpolation methods, K-nearest neighbor algorithms, and support vector regression methods, their method demonstrated optimal data repair effectiveness, providing important technical support for ETC data quality control.

The establishment of anomaly data identification and processing mechanisms is crucial for ensuring ETC data reliability. Through classifying and identifying missing data types, systematic data repair approaches can achieve significant improvements in data quality. By setting reasonable business rules and statistical thresholds, problematic data such as speed anomalies, time interval anomalies, and path anomalies can be effectively identified through automated systems.

The construction of data integrity verification systems ensures the logical consistency of ETC data. By checking the matching of vehicle entry and exit records, verifying temporal sequence continuity, and validating accounting information accuracy, multi-level data verification mechanisms have been established. This system not only identifies data problems but also provides a basis for data repair, thereby ensuring transaction record completeness and traceability.

From a technological development perspective, ETC data quality control is evolving in intelligent and automated directions. Applications of emerging technologies, such as machine learning-based anomaly detection algorithms, deep learning-based data repair methods, and knowledge graph-based consistency verification technologies, are continuously improving data quality control levels.

3.1.3. Contradictory Findings and Boundary Conditions

A critical examination of the reviewed ETC literature reveals several contradictory findings regarding system performance that warrant a careful analysis to understand their underlying causes and boundary conditions. These apparent contradictions often reflect differences in operational contexts, methodological approaches, and application scenarios rather than fundamental technological inconsistencies.

The literature presents mixed findings regarding ETC data reliability across different traffic conditions. Weng et al. successfully demonstrated systematic information extraction technologies based on ETC electronic toll data for road segment travel speed calculation and traffic flow distribution analysis under highway conditions [16]. However, practical implementations have revealed varying performance levels depending on traffic density, road types, and environmental conditions. The studies by Xue et al. on highway traffic flow data repair and real-time prediction based on ETC toll data showed that data quality control becomes more challenging under certain traffic scenarios, requiring sophisticated preprocessing techniques [17].

Multi-source data fusion approaches also show performance variations that initially appear contradictory. While some studies report significant improvements through ETC data integration with other sensing technologies, others show more modest gains. These differences typically reflect the quality and complementarity of the additional data sources, the sophistication of the fusion algorithms employed, and the specific application requirements. The effectiveness of fusion approaches varies substantially based on the specific combination of technologies and operational contexts examined.

3.2. Traffic State Feature Extraction Methods

Traffic feature extraction technology based on ETC data embodies the transformation process from data to information and from information to knowledge. This process requires not only advanced algorithmic support but also a deep understanding of traffic operation patterns. The multi-level nature of traffic state feature extraction reflects both the complexity of transportation systems and the richness of ETC data, with each level from microscopic individual vehicle behavior to macroscopic network operations having its unique characteristics and value.

3.2.1. Microscopic Feature Extraction

Microscopic feature extraction focuses on individual vehicle behavioral characteristics, which serve not only as the foundation for macroscopic traffic states but also as important windows for understanding traffic operation mechanisms. The accurate extraction of microscopic features directly affects the reliability and effectiveness of subsequent analyses.

The precision and real-time capability of vehicle speed estimation technology constitute core requirements for microscopic feature extraction. The individual vehicle speed is calculated through the distance between adjacent ETC gantries and passage time, with the basic formula v = L/(t₂ − t₁), where L represents road segment length, and t₁, t₂ represent the moments when vehicles pass the front and rear gantries, respectively. Behind this seemingly simple calculation lies complex technical considerations, including the precise measurement of inter-gantry distances, timing synchronization accuracy, and communication delay compensation. In practical applications, technical details such as specific vehicle positions within gantry detection zones, detection trigger time differences, and vehicle matching in multi-lane situations must also be considered.

The representativeness of ETC speed measurements: An important consideration in ETC-based speed estimation is whether speeds measured at gantry locations accurately represent general traffic flow conditions. As described in Section 2.1, ETC systems utilize Dedicated Short-Range Communication (DSRC) technology to complete the toll collection process through wireless data communication between vehicles and roadside units. The system is designed to achieve “non-stop automatic charging,” meaning vehicles should be able to pass through ETC zones while maintaining normal operating speeds.

However, the representativeness of ETC speed measurements requires careful consideration: Driver behavior effects may occur as vehicles approach ETC gantries, particularly in areas where the infrastructure is clearly visible or during the initial phases of system deployment when drivers are still adapting to the technology. Measurement location factors should be considered, as ETC gantries provide point-specific speed measurements rather than continuous speed profiles along road segments. The measured speeds reflect conditions at specific cross-sectional locations and may not capture speed variations occurring between detection points.

Infrastructure design considerations can help mitigate potential representativeness issues: Non-intrusive detection methods employed in modern ETC systems, as outlined in the system architecture described in Section 2.1, minimize physical disruption to traffic flow. The use of electronic communication rather than physical barriers allows vehicles to maintain normal speeds during data collection. The strategic deployment of ETC gantries at locations representative of typical road segment conditions, rather than at points immediately adjacent to geometric constraints or traffic control devices, can improve the representativeness of collected data.

Multi-point measurement approaches can enhance the reliability of ETC speed data by utilizing multiple gantries along road segments to capture speed variations and validate measurement consistency. Data validation techniques can be employed to identify and filter anomalous speed readings that may result from unusual driver behavior or system malfunctions, as discussed in the data quality control methods outlined in Section 3.1.2.

While these considerations are important for the proper interpretation of ETC speed data, the extensive deployment of ETC infrastructure and the system’s design for free-flow operation provide a valuable foundation for traffic speed monitoring when appropriately implemented and interpreted.

The accuracy of travel time calculation methods directly affects the reliability of traffic state assessment. The multi-perspective travel time prediction model based on ETC data developed by Luo et al. demonstrates the technological frontier in this field [18]. The precision of traffic flow statistics technology forms an important foundation for traffic management decisions. By counting vehicles passing through ETC gantries within unit time periods, precise vehicle-type-specific flow data can be obtained. Compared with traditional detectors, ETC data can not only provide accurate vehicle counts but also distinguish different vehicle types and identify specific vehicles, enabling refined traffic analysis. More importantly, the vehicle identity information in ETC data makes flow statistics traceable, which holds significant value for traffic behavior analysis and policy effect evaluation.

From technological development trends, microscopic feature extraction is evolving toward higher precision, more dimensions, and greater intelligence. The applications of new technologies such as artificial intelligence-based feature extraction algorithms, edge computing-based real-time processing technologies, and multi-source fusion-based precision improvement methods are continuously advancing microscopic feature extraction capabilities.

3.2.2. Mesoscopic Feature Extraction

Mesoscopic feature extraction focuses on traffic states at the road segment and regional levels. Feature extraction at this level requires the comprehensive consideration of interactions among multiple microscopic elements, reflecting the dynamic characteristics and complexity of transportation systems.

The accuracy of road segment congestion identification technology serves as a prerequisite for traffic management responses. Yan Yongfei et al. proposed recurrent traffic congestion management strategies for urban expressways integrated with ETC systems, analyzing the characteristics of recurrent traffic congestion on urban expressways [19]. Based on ETC system functions and ramp control principles, they proposed expressway ETC ramp control and ramp entrance–exit OD flow analysis models. Through determining and analyzing key ramp entrances and exits and their different causes of recurrent congestion, they developed corresponding congestion management strategies, providing important references for expressway recurrent congestion management and ramp control.

The practicality of queue length estimation methods reflects the value of ETC data in refined traffic management. Ban et al. developed methods for real-time queue length estimation at signalized intersections using travel times from mobile sensors [20]. Their approach was based on the observation that critical pattern changes in intersection travel times or delays, such as discontinuities (sudden and dramatic increases in travel times) and non-smoothness (changes in slopes of travel times), indicate signal timing or queue length changes. By detecting these critical points in intersection travel times, real-time queue lengths can be reconstructed. They introduced the concept of “Queue Rear No-delay Arrival Time” and demonstrated how measured intersection travel times from mobile sensors can be processed to generate sample vehicle queuing delays, providing important support for signal control optimization.

Traffic density analysis technology is based on the classical flow–speed–density fundamental relationship, deriving traffic density from flow and speed data. Although this method has a mature theoretical foundation, practical applications must consider the characteristics and limitations of ETC data. Unlike continuous detection data, ETC data possesses discrete characteristics, requiring appropriate interpolation and fitting methods to estimate continuous density distributions. Additionally, factors such as equivalent conversion for different vehicle types, the effects of mixed traffic flows, and the representativeness of detection cross-sections must be considered.

Development trends in mesoscopic feature extraction are reflected in algorithmic intelligence, real-time processing, and result precision. The applications of new methods such as machine learning-based congestion pattern recognition, deep learning-based traffic state prediction, and reinforcement learning-based dynamic threshold optimization are continuously advancing mesoscopic feature extraction technologies.

To provide a comprehensive comparison of different feature extraction approaches,

Table 2. Comparison of ETC data-based traffic state feature extraction methods.

Analysis Level	Method	Primary Features	Data Requirements	Reported Performance	Real-Time Capability	Computational Complexity	Advantages	Limitations	Best Application Scenarios
Microscopic	Speed Estimation	Individual vehicle speed	ETC gantry timestamps and distances	Reliable for highway conditions	Excellent	Low	Simple calculation and continuous monitoring	Point-based measurement, affected by gantry spacing	Highway segments and arterial roads
	Travel Time Calculation	End-to-end travel time	Entry/exit timestamps	High precision for complete trips	Excellent	Low	Direct measurement and trackable vehicles	Requires complete trip data and limited to ETC users	Route planning and performance monitoring
	Vehicle Classification	Vehicle-type identification	OBU vehicle info and transaction data	High reliability for registered vehicles	Excellent	Low	Automated identification and real-time processing	Limited to ETC-equipped vehicles	Toll collection and fleet management
Mesoscopic	Congestion Detection	Traffic density and speed patterns	Multi-gantry speed data	Effective for recurrent congestion	Good	Medium	Spatial coverage and pattern recognition	Threshold dependency and calibration needs	Urban expressways and corridor management
	Queue Length Estimation	Queue formation patterns	Upstream/downstream delays	Moderate precision but requires validation	Good	Medium	Dynamic estimation capability	Requires model calibration and weather sensitivity	Intersection management
	Flow Distribution Analysis	Traffic volume patterns	Multi-point flow counts	Good for trend analysis	Good	Medium	Network-wide coverage and temporal analysis	Sample size dependency	Network planning and demand analysis
Macroscopic	Network State Assessment	Overall network performance	System-wide ETC data	Suitable for policy evaluation	Fair	High	Comprehensive network view	Computationally intensive and aggregation effects	Regional traffic management
	Origin–Destination Analysis	Trip patterns and demand flows	Complete trip records	Rich insights for planning	Poor	High	Detailed behavioral analysis	Privacy concerns and processing complexity	Transportation planning
	Reliability Assessment	Service quality metrics	Long-term historical data	Valuable for infrastructure assessment	Poor	High	Long-term trend analysis	Requires extensive historical data	Infrastructure investment decisions

presents a systematic comparison of ETC data-based traffic state feature extraction methods across microscopic, mesoscopic, and macroscopic levels. This comparison highlights the strengths, limitations, and applicability of various approaches, enabling researchers and practitioners to select the most appropriate methods for their specific applications.

3.3. Multi-Source Data Fusion Framework

Multi-source data fusion represents an advanced form of ETC data applications, embodying development trends in modern traffic sensing systems. The limitations of single data sources are increasingly apparent, while multi-source fusion can fully leverage the complementary advantages of different data sources, achieving simultaneous improvements in sensing precision and coverage. Multi-source fusion represents not only innovation in technical methods but also conceptual innovation in traffic sensing.

Technical challenges in data-level fusion primarily manifest in data format standardization, spatiotemporal reference unification, and quality control consistency. Different data sources employ different coordinate systems, temporal references, and precision standards, requiring effective data fusion through technical means such as coordinate transformation, temporal synchronization, and precision matching. Simultaneously, unified data quality evaluation systems must be established to ensure the reliability and consistency of the fused data.

Key technologies in feature-level fusion include feature selection, weight allocation, and fusion algorithms. Feature selection involves identifying the most relevant traffic parameters from each data source, such as precise vehicle identification capabilities from ETC systems and continuous trajectory information from floating car data. Weight allocation determines the relative contribution of different data sources in the fusion process, requiring the consideration of each source’s accuracy, coverage, and reliability characteristics.

Tang Keshuang et al. conducted simulation-based evaluation research on traffic state estimation accuracy using floating car data in complex road networks [21]. Through 30 simulation experiments using VISSIM traffic simulation software and taking the microscopic simulation network of Shanghai’s Lujiazui area as an example, they analyzed the effects of floating car proportions and data sampling frequencies on network coverage and average travel speed estimation accuracy. Their results demonstrated that as floating car proportions and sampling frequencies increase, average travel speed estimation accuracy and network coverage gradually improve, with optimal results achieved when floating car proportions reach 8% and sampling frequencies reach 1/10 s. This research provides empirical guidance for determining optimal fusion parameters in multi-source data integration strategies.

Fusion algorithms combine weighted features from different sources using various technical approaches, including statistical methods, machine learning techniques, and optimization algorithms. The effectiveness of fusion depends on proper algorithm selection based on data characteristics and application requirements. Factors such as feature timeliness, correlation, and independence must be considered to ensure the scientific validity and practicality of the fusion results. The research demonstrates that systematic approaches to multi-source data fusion can significantly improve traffic state estimation accuracy compared to single-source methods.

Decision-level fusion strategies synthesize the analysis results from multi-source data, construct data view models, and achieve a comprehensive assessment of urban road traffic problems. Fusion at this level no longer focuses on specific data and features but on the analysis results and decision support. By establishing scientific decision fusion mechanisms, the advantages of different analysis methods can be fully utilized to improve the scientific validity and accuracy of decisions. The implementation of decision-level fusion requires the establishment of comprehensive knowledge bases and rule bases, constructing intelligent decision support systems through combinations of expert experience and machine learning. Additionally, requirements for decision timeliness, accuracy, and interpretability must be considered to ensure the practicality and reliability of fused decisions.

From development trends, multi-source data fusion is evolving toward higher levels, deeper degrees, and broader scopes. The applications of new technologies such as edge computing, cloud computing, and artificial intelligence are providing stronger technical support for multi-source fusion. Simultaneously, trends toward standardization, platformization, and service orientation are driving multi-source fusion from technical research toward industrial applications.

A comprehensive analysis of the development status of ETC data-based traffic sensing and feature extraction technologies reveals that this field has formed relatively complete technical systems, achieving important progress in data processing, feature extraction, and fusion analysis. However, facing future development requirements, continued in-depth research in algorithm innovation, system integration, and application expansion remains necessary to achieve the maximum utilization of ETC data value.

4. Traffic State Analysis and Assessment Using ETC Data

Having established the technical foundations for ETC data processing and feature extraction, this section focuses on how these capabilities enable comprehensive traffic state analysis and assessment.

Compared with traditional traffic state analysis methods, ETC data-based analysis approaches possess unique advantages, including extensive coverage, strong temporal continuity, and trackable vehicle identity, providing robust technical support for refined urban traffic management.

4.1. Cross-Section Traffic State Analysis

Cross-sections serve as fundamental units for traffic flow analysis, and the accuracy of their state analysis directly affects the effectiveness of the entire traffic management system. ETC data-based cross-section traffic state analysis can not only replace traditional methods but more importantly achieves improvements in analysis precision, timeliness, and scalability. These technological advantages stem from the structured characteristics and high-quality assurance of ETC data, providing a solid foundation for constructing intelligent traffic state analysis systems.

Clustering-based cross-section classification methods embody the core concept of data-driven traffic analysis. By utilizing clustering algorithms such as K-means and fuzzy C-means to classify cross-sections based on characteristics including flow, speed, and occupancy, the objective quantification and scientific classification of traffic states can be achieved. Lv Nengchao et al. proposed travel pattern analysis and differentiated pricing strategies based on highway transaction data, employing self-organizing mapping algorithms to classify user characteristics [22]. The innovation of this research lies not only in focusing on the physical characteristics of traffic flow but also in deeply exploring user behavioral characteristics, providing theoretical foundations for personalized traffic management.

Li Junyi et al. investigated ETC data-based methods for classifying passenger car commuting and tourism groups [23]. This research employed data analysis methods, including association analysis and cluster analysis, to conduct the multi-faceted characteristic and threshold analyses of user groups, constructing a systematic process framework for distinguishing passenger car commuting groups from tourism groups. The significant contribution of this research lies in organically combining traffic state analysis with user behavior analysis, providing technical support for precise traffic services. The research validated the effectiveness of the classification methods, demonstrating that the adopted analytical approaches can accurately explore the potential and value of passenger car commuting and tourism groups, offering theoretical support for ETC enterprise business expansion. This traffic state analysis method, based on user group characteristics, not only improves analysis specificity and accuracy but also provides scientific foundations for formulating differentiated traffic management strategies.

Zhao Jing et al. proposed fatigue driving identification methods for ETC intelligent systems [24]. This research addressed the driving traffic safety behaviors of fatigued drivers by proposing application research on fatigue driving identification methods based on ETC intelligent systems at service areas. Building upon the full utilization of existing ETC gantry infrastructure, the research achieved the automatic collection and analysis of vehicle driving records through deploying ETC antennas at highway service area entrances and exits, providing warning information to ETC vehicles traveling on highways. The innovation of this research lies in organically combining traffic safety management with ETC systems, improving the accuracy and timeliness of traffic safety warnings through intelligent and informatized technology applications. The cloud platform analyzes vehicle driving records from vehicle passage data uploaded by ETC antennas, thereby achieving the intelligent identification and proactive warning of fatigue driving behaviors. This technical approach not only improves traffic safety management efficiency but also provides referenceable technical pathways for similar intelligent applications.

4.2. Road Segment Traffic State Analysis

Road segments serve as fundamental units connecting cross-sections, and their traffic state analysis complexity far exceeds single-point cross-section analysis. Segment analysis requires the comprehensive consideration of multiple-dimensional factors, including spatial continuity, temporal dynamics, and flow variability, placing higher demands on the systematicity and precision of analytical methods. ETC data-based road segment traffic state analysis technology achieves the comprehensive sensing and precise assessment of segment traffic states through fully utilizing the spatiotemporal continuity characteristics of the data.

The construction of road segment operational reliability assessment systems reflects an important shift in traffic management from focusing on the average levels to emphasizing service quality. Traditional traffic analysis often concentrates on indicators such as average speed and average delay, while reliability assessment focuses more on service stability and predictability.

Calculating these indicators based on ETC data possesses advantages, including large data volumes, extended temporal spans, and high precision. Compared with traditional floating car surveys, ETC data can provide larger sample sizes of travel time data, making reliability indicator calculations more accurate and stable. Simultaneously, the long-term continuity characteristics of ETC data enable the analysis of reliability variation patterns under different time periods, weather conditions, and traffic situations, providing more scientific foundations for traffic management strategy formulation.

The precision of bottleneck segment identification technology constitutes a crucial component of traffic optimization management. By analyzing the spatial distribution and temporal evolution of segment speeds to identify bottleneck locations causing congestion, the effectiveness of this technical method depends on the spatiotemporal resolution of the data and the scientific validity of the analytical algorithms. Bottlenecks typically manifest as the locations of sharp speed decreases, but accurate bottleneck identification requires the consideration of comprehensive influences from multiple factors. The advantage of ETC data in bottleneck identification lies in providing precise vehicle trajectory information. By analyzing numerous vehicle driving trajectories, the specific locations and times of speed decreases can be accurately pinpointed. Compared with traditional fixed detectors, ETC data can provide more continuous and complete speed variation processes, making bottleneck identification more accurate and reliable. The practical application significance of bottleneck identification lies in providing scientific foundations for traffic organization optimization. By identifying characteristics such as bottleneck locations, occurrence times, duration, and impact scope, targeted improvement measures can be formulated, including signal optimization, lane management, and traffic guidance, thereby improving overall segment traffic efficiency.

4.3. Regional Traffic State Analysis

Regional-level traffic state analysis requires the comprehensive consideration of interactions among multiple road segments and nodes, with complexity and systematicity far exceeding the analysis of individual cross-sections or segments. Regional analysis not only requires processing large volumes of multi-source heterogeneous data but also demands an understanding of complex spatial correlation relationships and temporal evolution patterns. ETC data-based regional traffic state analysis technology achieves the comprehensive sensing and scientific assessment of regional traffic operational states by constructing systematic analytical frameworks.

Regional congestion situation index construction represents the core technology for regional traffic state analysis. Constructing comprehensive indices reflecting overall regional congestion levels based on ETC data requires the comprehensive consideration of multi-dimensional traffic parameters and the establishment of scientific index calculation models. Peng Wuxiong and Li Jianzhong constructed traffic operation analysis chromatograms based on Wuhan’s ETC system [25], providing important technical references for regional traffic state visualization analysis. This research was founded on speed–flow models, comprehensively considering flow and speed indicators based on floating car and ETC river-crossing flow data, separately determining their classification standards. Based on this foundation, traffic operation analysis chromatograms were constructed, avoiding potential operational status misjudgments that might occur when employing single flow indicators. The application results from Wuhan demonstrate that these chromatograms can provide intuitive and powerful technical support for the real-time monitoring of road operational conditions, analyzing road anomalies, and quantitatively analyzing actual road capacity and free-flow speeds. The innovation of this research lies in achieving the intuitive expression of complex traffic states through chromatographic representation, enabling traffic managers to rapidly understand the overall regional traffic operational conditions. Compared with traditional digital indicators, chromatograms possess advantages, including strong intuitiveness, ease of understanding, and convenience for comparison, providing more user-friendly technical interfaces for traffic management decision-making.

Inter-regional correlation analysis reveals inherent connections and mutual influences within urban transportation systems. Analyzing traffic flow correlations among different regions and identifying the source–sink relationships of traffic flows holds significant importance for understanding the overall operational mechanisms of urban transportation. Inter-regional correlations manifest not only in quantitative relationships of traffic flows but also in temporal synchronization, spatial continuity, and influence transmission across multiple dimensions. The vehicle identity tracking characteristics of ETC data provide unique data foundations for inter-regional correlation analysis. By analyzing vehicle flow trajectories among different regions, traffic connection intensity and directional characteristics between regions can be accurately identified. This analysis can not only reveal the spatial structural characteristics of urban transportation but also predict the transmission effects of traffic state changes between regions. The application value of inter-regional correlation analysis lies in providing scientific foundations for regional coordinated management. By understanding mutual influence relationships among different regions, coordinated and unified management strategies can be formulated, avoiding the poor overall effects resulting from local optimization and achieving the systematic optimization of regional traffic management.

A comprehensive analysis of the development status of ETC data-based urban traffic state analysis and assessment technologies reveals that this technical system has formed relatively complete analytical frameworks across three levels: cross-sections, road segments, and regions. Based on technological development trends, future research priorities will concentrate on intelligent analytical methods, precise analytical results, and real-time analytical applications. The applications of new technologies, including artificial intelligence, big data, and cloud computing, will further promote technological innovation and application expansion in this field.

5. ETC-Based Urban Traffic Management: Applications and Practices

5.1. Real-Time Monitoring and Signal Control Optimization

Real-time monitoring systems based on ETC data have been implemented in multiple cities. Yu Xinhai investigated the comprehensive applications of ETC gantry big data based on cloud computing, proposing a two-tier cloud computing architecture of “edge cloud + central cloud.” This architecture leverages the virtualization, large-scale integration, high reliability, and high scalability characteristics of cloud computing technology to achieve the asset-based management of ETC gantry data for comprehensive solutions, including precise billing, vehicle-road cooperation, thematic analysis, traffic modeling, real-time monitoring systems, and intelligent operation and maintenance systems [26]. This architecture provides robust technical support for urban traffic real-time monitoring, enabling second-level traffic state updates and responses.

In signal control optimization, Feng et al. proposed real-time adaptive signal control methods for connected vehicle environments, solving phase sequences and durations through two-level optimization problems, considering two objective functions: total vehicle delay minimization and queue length minimization. Research demonstrates that under high-penetration conditions, this control algorithm can reduce the total delay by 16.33% compared to traditional actuated control, while maintaining similar delay levels under low-penetration conditions [27]. Wang Wenwen investigated ETC-based traffic flow collection methods and intersection cooperative control models, proposing a traffic model for adjacent intersection cooperative control (ICC model). This model combines the remaining green time from upstream and downstream intersections, considers multiple signal states when vehicles arrive, and significantly improves lane vehicle traffic efficiency through the reasonable matching and setting of phase differences between adjacent signalized intersections [28]. He et al. developed PAMSCOD, a queue-based arterial multi-modal signal control system that executes arterial traffic signal control in vehicle-to-infrastructure communication environments while considering multiple travel modes. This system can significantly reduce delays for both transit and automobiles under both saturated and oversaturated traffic conditions [29].

Regarding signal control in intelligent connected vehicle environments, Chen and Englund conducted a comprehensive review of cooperative intersection management, discussing major technologies and solutions for both signalized and non-signalized intersections, providing technology roadmaps for the future development of automated and cooperative intersections [30]. Guo et al. systematically reviewed urban traffic signal control in connected and automated vehicle environments, reviewing the methods and models for estimating traffic flow states and optimizing traffic signal timing schemes based on CAV. They classified CAV-based traffic control methods into six types and proposed conceptual mathematical frameworks that can be specified for each method through selecting different state variables, control inputs, and environmental inputs [31]. Guo et al. further proposed joint optimization methods for vehicle trajectories and intersection controllers for connected automated vehicles, employing algorithms combining dynamic programming and shooting heuristics (DP-SH). This approach can simultaneously optimize CAV trajectories and intersection controllers (traffic signal timing and phasing), reducing average travel time by 35.72% and fuel consumption by 31.5% compared to adaptive signal control [32].

These studies demonstrate that ETC data-based real-time monitoring and signal control optimization technologies have transitioned from traditional passive-response modes to proactive-prediction and intelligent-control modes. These technologies not only significantly improve traffic efficiency but also provide important technical reserves and application foundations for urban traffic management in future intelligent connected vehicle environments.

5.2. Traffic Prediction and Artificial Intelligence Technology

Artificial intelligence technology plays an increasingly important role in ETC data-based traffic prediction, becoming a key technology for enhancing the intelligence levels of urban traffic management. Jiang et al. proposed highway traffic parameter prediction models based on combined deep learning models, constructing two deep learning time-series combination prediction models. Using ETC data from the Shanlin Highway in Gansu Province for short-term prediction validation, the results demonstrated that both models exhibited ideal data training and prediction capabilities in testing [33].

In traffic state estimation, Zhao and Yu proposed boundary-aware observer–deep learning traffic state estimation methods that integrate partial differential equation observers with deep learning paradigms, estimating spatiotemporal traffic states from boundary sensing data [34]. Zhao et al. developed an improved support tensor machine method for traffic state subcategory classification based on ETC gantry data (ISTM). Compared with K-means clustering and SVM, ISTM demonstrated optimal values for both SumD and DBI indicators [35].

Regarding comprehensive reviews of deep learning algorithms, Afandizadeh et al. conducted comprehensive reviews of deep learning algorithms in traffic prediction, covering different scenarios across various traffic datasets and reviewing 111 pioneering research works since the 1980s, including both deep learning and classical models. The research provided detailed introductions to the data sources used in transportation systems and in-depth investigations of theoretical foundations for popular deep learning algorithms and classical models in traffic prediction [36]. In explainable artificial intelligence, Guo et al. proposed explainable traffic flow prediction methods based on large language models (xTP-LLMs), capturing complex time-series patterns and external factors by converting multi-modal traffic data into natural language descriptions. Experiments demonstrated that the xTP-LLM provides intuitive and reliable explanations for predictions while maintaining accuracy comparable to deep learning baselines [37].

Shaygan et al. provided comprehensive reviews of the recent advances and emerging opportunities for artificial intelligence in traffic prediction, focusing on AI-based traffic prediction methods with a particular emphasis on multivariate traffic time-series modeling. The research covered various data types and resources, classified fundamental data preprocessing methods in traffic prediction contexts, subsequently summarized prediction methods and applications, provided solutions for major research challenges in traffic prediction, and discussed future research directions [38]. Liu and Shin reviewed traffic flow prediction methods in intelligent transportation system construction, classifying methods into three categories based on statistics, machine learning, and deep learning, concluding that deep learning methods achieved optimal overall effectiveness [39]. Han conducted traffic volume prediction based on deep learning models for Annual Average Daily Traffic (AADT) correction, managing data discontinuity through LSTM technology [40].

In video data processing, Cai et al. developed high-precision deep learning models for estimating traffic characteristics from video data. Through deep learning video processing algorithms, they detected, tracked, and predicted highly disaggregated vehicle-based data such as trajectories and speeds, converting these data into aggregated traffic characteristics, including speed variance, average speed, and flow. Using LiDAR sensor observations as the ground truth, the results demonstrated that this technology estimates lane traffic volume with 97% accuracy, while computer vision technology can estimate individual vehicle-based speed calculations with 90–95% accuracy [41].

Recent advances in graph neural networks (GNNs) have demonstrated a significant potential for enhancing ETC-based traffic analysis through sophisticated spatiotemporal modeling approaches. These developments address critical challenges in capturing complex relationships within transportation networks that are highly relevant to ETC system applications.

The integration of attention mechanisms with graph structures has shown particular promise. Geng et al. introduced STGAFormer, a spatial–temporal gated attention transformer that enhances long-term prediction capabilities and handles sudden traffic incidents through improved temporal feature extraction [42]. This approach is directly applicable to ETC systems, where gantry networks can benefit from attention-based mechanisms to focus on relevant spatial relationships during varying traffic conditions, particularly valuable for managing unexpected events that affect traffic flow patterns captured by ETC data.

Multi-dimensional temporal pattern recognition has emerged as another crucial advancement. Ju et al. proposed COOL, a conjoint spatiotemporal graph neural network that models high-order spatiotemporal relationships through heterogeneous graph construction [43]. The conjoint learning framework aligns well with ETC data characteristics, where spatial relationships between the gantries and temporal sequences of vehicle passages naturally form complex graph structures, requiring sophisticated modeling approaches.

For multi-modal transportation applications, Baghbani et al. developed the TMS-GNN, which specifically addresses traffic-aware multistep prediction while considering sensitivity to urban traffic conditions [44]. This methodology offers valuable insights for ETC-based transit management, where bus networks’ coexistence with general traffic mirrors the operational environment of urban ETC systems, suggesting potential applications in integrated ETC–transit optimization strategies.

These GNN advancements provide theoretical foundations and technical approaches that can enhance ETC data analysis capabilities, particularly in modeling complex network relationships, handling temporal dependencies, and integrating multi-source traffic information within urban transportation systems.

Overall, although some studies do not directly utilize ETC data, their approaches and methods provide valuable references. Simultaneously, artificial intelligence technology applications in ETC data traffic prediction have evolved from simple statistical analysis to complex deep learning models, from single-modal data processing to multi-modal fusion analysis, and from black-box prediction to explainable artificial intelligence. These technological advances provide robust technical support for the future intelligent and precise applications of ETC data fusion in urban traffic management.

5.3. Transformation of Highway Technology to Urban Applications

Optimization technologies for highway ETC systems provide important technical references and application foundations for traffic management on urban expressways and arterial roads. Through appropriate adjustments and optimization, these technologies can be effectively applied in urban traffic environments. Zou et al. proposed dynamic generation methods for highway ETC gantry topology based on LightGBM, using ETC gantry and toll station transaction data from a province in southeastern China to dynamically update ETC gantry topology data through machine learning algorithms. Comparisons with 14 other machine learning algorithms demonstrated that LightGBM performed optimally with 97.6% accuracy. This method can be directly applied to ETC gantry layout optimization for urban expressways [45].

Zhang et al. proposed data-driven optimization-based highway traffic state estimation methods, fusing ETC data with detector data and leveraging the extensive coverage provided by ETC data and fine-grained characteristics provided by detector data. By capturing probabilistic interdependencies between traffic state segments and their corresponding upstream and downstream segments, they developed two optimization models based on maximum likelihood and maximin likelihood principles. In case studies on the Zhejiang G92 Highway, the two optimization models achieved mean absolute percentage errors of 0.9% and 2.3%, respectively, during peak hours. This multi-source data fusion approach can be fully applied to traffic state estimation for urban expressways and arterial roads [46].

Kim et al. proposed methods for dynamically determining toll plaza capacity through the real-time monitoring of approaching traffic conditions. Applications at San Francisco Bay Area toll plazas demonstrated that vehicle storage quantity estimates based on discharge rates and travel times can serve as proxy measures for predicting toll plaza capacity change effects. This dynamic capacity adjustment method is equally applicable to the management of urban toll road segments and expressway entrances [47].

In traffic demand understanding, Kim et al. utilized electronic toll collection data to understand traffic demand, conducting dynamic origin–destination traffic volume estimation based on Hanshin Expressway ETC data. Research demonstrated that ETC data facilitates an understanding of traffic demand and its variations, clarifying the influences of external factors on OD flow changes and proposing statistical analysis methods for variations based on these factors. Results can be broadly applied to urban network management [48]. Weng et al. investigated traffic information mining technologies based on ETC data, conducting a case analysis using Beijing’s 2008 ETC data as examples. They proposed highway travel speed calculation models, traffic flow distribution analysis methods, and methods for obtaining indicators such as origin–destination distributions and vehicle-type proportion distributions. These data processing and extraction technologies are equally applicable to urban road traffic operations and planning [16].

Chen and Chen explored the spatiotemporal mobility of highway traffic flows based on ETC data for precise travel time estimation and prediction, proposing a travel time estimation and prediction (TTEP) framework. Through Weighted Root-Mean-Square Similarity (Weighted-RMSS) methods and Multi-Slope-Based Linear Regression (Multi-SBLR) methods, they improved the prediction accuracy for highway travel times. Experimental results demonstrated that this approach outperformed existing methods and could significantly reduce prediction errors for highway travel times. This framework can be applied to travel time prediction for urban arterials and expressways after appropriate adjustments [49].

In toll plaza management optimization, Wang et al. analyzed the lane capacity of highway toll plazas using toll data. Targeting the coexistence of electronic toll collection lanes and compound toll collection lanes at Chinese highway toll plazas, they employed real toll data-driven methods to analyze the capacity of toll plaza entrance and exit lanes under mixed traffic conditions. This method can be directly applied to capacity analysis for urban toll road segments and parking facilities [50]. Komada et al. investigated traffic flows on toll highways with electronic and traditional toll plazas, discovering that electronic and manual toll vehicles enter their respective lanes at low densities while mixing at each toll plaza at high densities. They derived fundamental diagrams for electronic and manual toll vehicles, with these fundamental diagram analysis methods equally applicable to urban mixed traffic flow research [51].

Choi et al. investigated traffic congestion dynamics using highway electronic toll collection system data, demonstrating how real Taiwan highway system data can monitor traffic congestion evolution on fundamental diagrams. They discovered that congestion characteristics form cyclic trajectories, with areas enclosed by loops increasing as congestion severity increases. The area enclosed by loops in fundamental diagrams serves as a measure of economic losses due to congestion. This congestion dynamics analysis method provides important theoretical foundations for congestion management on urban expressways and arterials [52].

It is noteworthy that although these studies are primarily based on highway environments, their core technologies and methodologies can be fully transferred to urban traffic environments. Scenarios such as urban expressways, arterial toll road segments, and large parking facilities share similar traffic characteristics with highways, enabling ETC systems to achieve comparable effectiveness in these scenarios. Particularly, in the current context of urban traffic digital transformation, these mature highway ETC technologies provide solutions and technical pathways for urban traffic management, facilitating the rapid enhancement of intelligent levels in urban traffic management.

5.4. ETC Technology Protocols and System Design

ETC system technology protocols and system design constitute the technical foundation for urban traffic applications. The sophistication of these underlying technologies directly affects the effectiveness and stability of upper-level applications. Zhang Hua’an conducted in-depth research on vehicle path recognition methods in intelligent transportation systems, separately investigating two different scenarios of highways and urban roads, with a primary focus on highway situations, while conducting a preliminary exploration of urban road scenarios. Against the background of achieving province-wide highway networked toll collection in Guangdong Province, addressing ETC vehicle path recognition problems on Guangdong highways, he proposed ETC vehicle path recognition methods based on ETC identification points, replacing manual identification stations with ETC identification point systems. This system’s implementation improved vehicle travel speeds, reduced operational costs, and generated favorable economic and social benefits [53].

Wang Chao conducted the research and design implementation of DSRC protocols in ETC systems, developing DSRC protocol implementation and design schemes oriented toward ETC application requirements to provide reliable communication support for ETC applications. The research combined ETC systems to conduct an in-depth analysis of DSRC protocol fundamental theories and related technical standards, analyzing the structures and characteristics of each protocol stack layer, establishing ARM-Linux software and hardware development environments, and conducting design implementation work for physical layers, data link layers, application layers, and device application layers. Through testing DSRC protocol software, design objectives were achieved, correctly implementing RSU and OBU communication functions specified by DSRC protocol standards with excellent stability [54]. Chai Long investigated the ETC lane system software design based on DSRC devices, designing system software for ETC lanes based on existing ETC non-stop toll collection system schemes in China. The research provided brief descriptions of technical backgrounds for ETC road toll collection systems, employing 5.8G frequency RFID technology, conducted a brief analysis of the information interaction processes among computers, Road-Side Units, and On-Board Units using DSRC protocol standards at highway toll stations in China, while also conducting a comparative analysis of communication interface schemes between lane computers and roadside units, as well as lane layout schemes [55].

In traffic control strategies, Papageorgiou provided an overview of road traffic control strategies, noting that traffic congestion leads to the severe deterioration of network infrastructure and correspondingly reduced throughput, which can be countered through appropriate control measures and strategies. The research provided brief discussions of future requirements for this important technical field [56]. Kato et al. proposed deep learning visions for heterogeneous network traffic control from deep learning perspectives, developing supervised deep neural network systems and describing how the proposed systems work and their differences from traditional neural networks. Preliminary reported results demonstrated that the proposed deep learning systems exhibited encouraging performance in signaling overhead, throughput, and delay compared to benchmark routing strategies (Open Shortest Path First (OSPF)) [57].

These fundamental technical research efforts establish solid technological foundations for widespread applications of ETC systems in urban transportation. From communication protocol standardization to system software optimization, from path recognition algorithm improvements to deep learning technology introduction, each technical component’s refinement directly affects the ETC system application’s effectiveness in urban traffic management. Particularly, the standardized implementation of DSRC protocols and the optimized design of lane system software provide important guarantees for stable ETC system operation in complex urban traffic environments.

5.5. Environmental Impact and Benefit Assessment

ETC system applications in urban transportation require not only a consideration of technical feasibility but also a comprehensive evaluation of environmental benefits and economic value, which holds significant importance for policy formulation and system promotion. Ramandanis et al. comprehensively assessed the environmental and economic footprints of electronic toll collection lanes through simulation studies, using toll stations in the Thessaloniki metropolitan area of Greece as case studies, considering specific traffic characteristics, variable toll station configurations, and different penetration rates of ETC tag users for automobiles and heavy vehicles. Results demonstrated that when toll station right lanes are converted to ETC lanes, private vehicle penetration rates must exceed 15% and heavy truck penetration rates must exceed 20% to reduce traffic congestion and improve environmental conditions [58].

Xing et al. conducted in-depth research on the safety assessment of toll plaza diverging areas, considering different vehicle toll collection types, noting that different vehicle toll collection types and different toll plaza distributions cause toll plaza diverging areas to become typical vehicle-weaving zones with frequent crossing behaviors and conflicts on highways. The research identified the influencing factors of four random parameters on conflict risks through developing random parameter ordered logit models. Results demonstrated that vehicle following patterns with identical toll collection types exhibit higher percentages of severe conflict risks, with an average acceleration of the following vehicles, lane marking indicators, initial vehicle lanes, and lane changes being significantly correlated with collision risk levels [59].

Regarding environmental pollution, Lai et al. specifically investigated the effects of manual and electronic toll collection systems on highway particulate pollutant levels in Taiwan, conducting two-week sampling activities at toll stations in the Taipei metropolitan area, collecting indoor toll station monitoring samples from ETC and MTC lanes to assess the hourly concentrations of polycyclic aromatic hydrocarbons, PM1.0, PM2.5, and PM10. Results demonstrated that automobile and truck lanes with MTC exhibited higher average concentrations of particulate polycyclic aromatic hydrocarbons, PM1.0, and PM2.5, compared to ETC. PM1.0 and PM2.5 concentration increases per 100 vehicles in bus and truck lanes with MTC were approximately twice those of ETC. ETC technology provides solutions for traffic delay problems at toll stations in Taiwan and may more uniformly reduce polycyclic aromatic hydrocarbons, PM1.0, and PM2.5 particulate pollutant levels from highway traffic emissions compared to MTC [60].

Vu conducted an in-depth exploration of global trends in highway electronic toll collection, Vietnam’s experience, and policy implications. By analyzing the global ETC trends and in-depth investigation of Vietnam’s experience, practical policy insights were provided for developing countries facing this critical transition. A key feature of the research was the in-depth analysis of Vietnam’s transition from manual toll collection to ETC, demonstrating significant improvements in emission reduction, operational efficiency, and cost and time savings. The research emphasized ETC’s role in improving productivity, promoting sustainable development, reducing pollution, and advancing toward an intelligent economic society, highlighting the critical importance of government leadership and private sector cooperation in achieving successful ETC implementation [61]. Al-Deek et al. provided detailed reports on the operational benefits of electronic toll collection through case studies, investigating traffic operational improvements at Orlando–Orange County Expressway Authority electronic toll plazas. Service times, vehicle arrival times, departure times, and vehicle count data were collected before and after installing the E-PASS automatic vehicle identification technology. Results demonstrated that for dedicated E-PASS lanes, the measured capacity increased threefold, the service time decreased by five seconds per vehicle, the average queuing delay decreased by one minute per vehicle, the maximum queuing delay decreased by 2.5–3 min per vehicle, and the total queuing delay for these lanes decreased by 8.5–9.5 vehicle-hours per morning peak hour [62]. Chauhan and Chauhan proposed intelligent toll collection systems for moving vehicles based on Indian conditions, proposing effective methods for locating vehicle license plates, creating databases, and linking to test prototype toll collection system performance for moving vehicles. Results demonstrated significant reductions in vehicle waiting times, queue lengths, fuel waste, and pollution emissions at toll stations, with future potential for theft control applications [63].

In economic benefit assessment, Chu et al. conducted a scenario analysis of cost–benefit evaluation for Taiwan’s ETC system, using mathematical and engineering economic methods to demonstrate the benefits and costs related to different stakeholders, simulating three traffic scenarios to investigate the parameters by changing assumptions about time savings, fuel savings, environmental improvements, and user costs, as well as outcome variations for different stakeholders. Research found that under current mechanisms, although ETC systems generate overall benefits greater than costs for society as a whole, they are not favorable for investors and highway bureaus [64].

These studies demonstrate that the environmental benefits and economic value of ETC systems have been fully validated. From reducing air pollution to improving traffic efficiency, from lowering operational costs to enhancing user experience, ETC systems exhibit significant positive effects across multiple dimensions. These research findings provide robust scientific foundations and policy support for promoting ETC system applications in urban transportation.

5.6. System Integration and Technology Development Trends

ETC system technological development is evolving in increasingly intelligent and integrated directions, with edge computing, system security, and multi-technology fusion becoming important driving forces for the deep development of ETC applications. Zhou et al. conducted an in-depth exploration of visions and challenges when intelligent transportation system sensing combines with edge computing, noting that the widespread use of mobile devices and sensors has driven the development of data-driven applications that can leverage big data power to benefit many aspects of our daily lives [65].

In system security, Avcı and Koca conducted a detailed analysis of intelligent transportation system technologies, challenges, and security issues, noting that protecting ITS infrastructure from cyber attacks has become a matter of national reputation. Providing the necessary technological infrastructure for the integrated operation of the systems used in ITS, particularly geographical location, communication, and mapping, is crucial. The research included the minimum security requirements that might be adopted for ITS applications and infrastructure against these attacks, as well as the most important cyber attacks that might occur in ITS applications [66].

In technology detection and improvement, Ho and Chung developed information-aided intelligent schemes for improving the vehicle flow detection of traffic microwave radar detectors, noting that to achieve satisfactory traffic management in intelligent transportation systems, traffic microwave radar detectors providing real-time traffic information with high precision are crucial [67]. In traffic flow model validation, Romanowska and Jamroz compared the traffic flow models with actual traffic data based on a quantitative assessment, proposing universal quantitative methods for evaluating the fundamental relationship models based on real traffic data from Polish highways. The proposed methods aimed to address the problems of finding optimal deterministic models to describe empirical relationships among fundamental traffic flow parameters based on simple transparent criteria [68].

Chang et al. investigated operational and design upgrades for electronic toll collection, examining effective applications of ETC technology in Taiwan highway and expressway planning and design applications, reviewing various design and operational issues, including current AVI/TOLL technology applications, potential operational enhancements, and alternative design upgrades, targeting effective electronic toll collection and electronic toll collection and traffic management [69]. Zhou et al. proposed GNSS-based electronic toll collection and road pricing systems, noting that ETC allows charging for road infrastructure use while minimizing traffic interference, achieving better social acceptance, and reducing costs. However, most road pricing strategies today are not based on satellite positioning. The research described the system architecture and operational principles of multi-service ETC systems based on Global Navigation Satellite Systems, the detailed design and functions of three subsystems, and provided a brief analysis of feasibility for GNSS-based road pricing in China [70].

Comprehensively, ETC system technological development has expanded from single toll collection functions to comprehensive traffic management platforms. Edge computing technology applications have made data processing more efficient, system security technology improvements have ensured stable system operation, and the development of multi-source data fusion and intelligent analysis technologies has enhanced system intelligence levels. These technological development trends indicate that ETC systems are becoming important components of urban intelligent transportation systems, providing robust technical support for constructing more intelligent, efficient, and safe urban traffic environments. Future ETC systems will achieve deeper system integration and functional expansion driven by emerging technologies such as 5G communication, artificial intelligence, and the Internet of Things, opening new pathways for the modernized development of urban traffic management.

6. Current Challenges and Future Research Directions

While the reviewed literature demonstrates significant progress in ETC-based traffic sensing technologies, several methodological and practical challenges emerge from this comprehensive analysis. The comparison tables presented in this survey reveal that no single approach excels in all scenarios, highlighting the need for context-specific technology selection and hybrid methodologies that can adapt to varying traffic conditions and application requirements.

Although ETC data-based urban traffic state sensing and analysis technologies have achieved significant progress in both theoretical research and practical applications, they still face numerous technical challenges and application bottlenecks in the process of developing toward larger-scale, deeper-level, and broader-scope applications. These challenges involve not only breakthrough requirements at the technical level but also concern the coordinated development of industrial ecosystems and comprehensive policy environment support. Meanwhile, with the continuous emergence of new technologies and evolving urban traffic demands, this field is experiencing unprecedented development opportunities and innovation space.

6.1. Technical Challenges

6.1.1. Data Processing Capabilities

With the comprehensive deployment of ETC systems nationwide and the continuous expansion of urban application scenarios, data volumes are experiencing explosive growth. The real-time processing of massive data has become a critical bottleneck, constraining system performance and application effectiveness.

Technical challenges related to real-time requirements manifest across multiple levels. Traffic management decisions require second-level or even millisecond-level data processing response capabilities, but traditional centralized data processing architectures struggle to meet these requirements. The introduction of edge computing technology provides new technical pathways for addressing this challenge. Zhou et al. noted that the combination of intelligent transportation system sensing with edge computing offers considerable benefits for transportation systems in terms of efficiency, security, and sustainability [65]. By shifting partial computing tasks to roadside devices and edge nodes, data transmission delays can be significantly reduced, and system response speeds can be improved. However, edge computing deployment also introduces new technical challenges, including computing capability limitations of edge devices, distributed data consistency assurance, and edge-cloud collaborative scheduling optimization, requiring in-depth research in system architecture design, algorithm optimization, and resource management.

Algorithm efficiency optimization represents another critical pathway for enhancing data processing capabilities. Traditional batch processing algorithms can no longer accommodate real-time processing requirements for streaming data, necessitating the development of efficient algorithms specifically designed for streaming data. These algorithms must not only achieve an optimal balance between accuracy and real-time performance but also possess excellent scalability and fault tolerance. Deep learning algorithms demonstrate enormous potential in traffic data processing, but their characteristics of high computational complexity and substantial resource consumption place higher demands on system design. How to enhance algorithm efficiency through technical approaches such as model compression, quantization acceleration, and parallel computing, while ensuring algorithm effectiveness, represents an urgent technical challenge that requires resolution.

6.1.2. Privacy Protection and Data Security

ETC data contains substantial sensitive personal privacy information, including vehicle identity, travel trajectories, temporal habits, and location preferences. The leakage of this information could pose serious threats to user privacy. How to effectively protect user privacy while fully utilizing data value has become a key factor constraining widespread ETC data applications.

Although the development of data anonymization technologies has alleviated privacy protection pressures to some extent, achieving an optimal balance between anonymization intensity and data usability remains challenging. Traditional anonymization methods, such as data anonymization and generalization processing, often significantly reduce data analytical value, affecting data-based decision-making effectiveness. Therefore, privacy protection technologies require comprehensive breakthroughs in theoretical foundations, algorithm design, and system implementation.

Differential privacy protection technology provides new approaches for addressing this issue. By incorporating carefully designed noise in data publishing and analysis processes, individual privacy can be protected while maintaining the overall statistical characteristics of data. However, differential privacy technology still faces numerous challenges in ETC data applications, including the reasonable allocation of privacy budgets, the optimized setting of noise parameters, and the collaborative protection of multi-dimensional data. Particularly in application scenarios requiring complex spatiotemporal analysis, maintaining the usability of analysis results while ensuring privacy protection effectiveness requires further theoretical research and technological innovation.

The improvement of data security transmission and storage technologies remains equally crucial. With the deep integration of ETC systems with other intelligent transportation systems, data flow pathways and storage nodes have become increasingly complex, making traditional encryption technologies insufficient for full-chain security protection requirements. The applications of emerging cryptographic technologies, such as blockchain technology, homomorphic encryption, and secure multi-party computation, provide technical support for constructing more secure and reliable data processing environments. However, these technologies still face challenges in actual deployment, including substantial computational overhead, high system complexity, and low standardization levels, requiring comprehensive trade-offs in technological maturity, deployment feasibility, and cost-effectiveness.

6.1.3. System Integration and Standardization

The current situation of inconsistent ETC system construction standards and varied data formats across different regions seriously impedes data-sharing applications and collaborative system development. This lack of standardization not only affects technology promotion and application but also increases the complexity and costs of system maintenance and upgrades.

Interface standard unification serves as the fundamental prerequisite for achieving system interconnection and interoperability. Although basic standard unification has been achieved for highway ETC systems domestically, significant differences still exist in system standards among different manufacturers, regions, and applications in urban application scenarios. This lack of standardization has led to data silo phenomena, limiting the full realization of the ETC data value.

The absence of data format standardization further exacerbates system integration difficulties. Different systems employ different data structures, encoding methods, and transmission protocols, making data exchange and fusion extremely complex. Although the relevant standardization organizations have conducted substantial work in promoting data format standardization in recent years, standard implementation varies considerably in practical applications, with compatibility issues remaining prevalent. This not only increases system development and maintenance costs but also affects the rapid promotion and application of new technologies.

6.1.4. Understanding Performance Boundaries and Operational Contexts

The contradictory findings in the ETC literature highlight the critical need for a better understanding of performance boundaries and operational contexts. Future research should focus on the systematic characterization of the conditions under which ETC systems maintain optimal performance, including comprehensive empirical validation across diverse urban environments. This includes establishing clear guidelines for system configuration, data quality thresholds, and integration strategies tailored to specific operational contexts.

6.2. Application Challenges

6.2.1. Coverage and Penetration Rate Enhancement Challenges

Although ETC user numbers have achieved rapid growth under policy promotion, the coverage and penetration rates on urban roads still face numerous challenges. Unlike the relatively singular application scenarios of highways, the complexity and diversity of urban traffic environments present greater difficulties for ETC system promotion and application.

Unbalanced infrastructure construction represents the primary factor affecting coverage rates. Urban road networks are dense with numerous nodes, requiring the consideration of multiple factors, including road classifications, traffic volumes, technical feasibility, and construction costs when deploying ETC gantries and reading–writing devices. The current ETC infrastructure is primarily concentrated on highways and urban expressways, with limited coverage on ordinary urban roads. This coverage imbalance not only restricts the application scope of ETC data in urban traffic management but also affects user experience and acceptance.

Changing user habits requires lengthy adaptation processes. Compared to the rigid demands for highway tolls, ETC applications on urban roads are often optional, making user willingness for active usage a key factor affecting penetration rates. How to enhance user willingness and satisfaction through technical improvements, service optimization, and incentive measures represents an important challenge for promotion and application.

Immature business models also constrain ETC promotion and application in urban scenarios. Unlike highway tolls, which possess clear business models, profitability models for urban ETC applications remain unclear. Application scenarios such as parking fees, congestion charges, and environmental fees possess enormous market potential but still face uncertainties in policy environments, technological maturity, and user acceptance. This business model uncertainty affects social capital investment enthusiasm and constrains the rapid development of related technologies and applications.

6.2.2. User Acceptance and Trust Establishment

User concerns about ETC system privacy and security represent important factors affecting acceptance. With frequent data breach incidents and increasing public privacy awareness, user demands for personal information protection have become increasingly strong. Data collected by ETC systems, such as location information and travel habits, possess high sensitivity, with users demanding higher requirements for the transparency, security, and controllability of data usage.

Insufficient technical transparency exacerbates user concerns. Most current ETC systems lack sufficient transparency in data collection, processing, and usage phases, making it difficult for users to understand how their data is being used and impossible to effectively supervise and control data usage. ETC systems in some regions still experience problems with system stability, recognition accuracy, and response speed, with equipment failures, misidentification, and billing errors occurring periodically. This information asymmetry not only affects user trust levels in systems but may also trigger broader social controversies and policy risks.

6.2.3. Investment Return Assessment Complexity

ETC system construction requires substantial infrastructure investment and continuous operational maintenance input. How to scientifically assess economic and social benefits to provide support for investment decisions represents a complex, systematic engineering challenge.

The quantitative assessment of social benefits faces methodological challenges. ETC system applications not only generate direct economic benefits but also produce significant social benefits, including environmental benefits, safety benefits, and public convenience benefits. Research by Lai et al. demonstrated that ETC technology can significantly reduce particulate pollutant emissions compared to manual toll collection [60]. However, these social benefits are often difficult to accurately quantify in monetary terms, creating challenges for comprehensive benefit assessment. How to establish scientific social benefit assessment systems and accurately quantify the comprehensive value of ETC systems represents the key to improving the scientific validity of investment decisions.

Trade-offs between long-term benefits and short-term costs increase assessment difficulties. ETC system construction investments are primarily concentrated in early phases, while benefit realization often requires extended timeframes. Factors such as technology upgrades, equipment updates, and standard changes may also affect system lifecycles and investment returns. This temporal mismatch makes traditional investment return analysis methods insufficient for accurately reflecting the true value of ETC systems, necessitating the development of more suitable assessment methods and tools.

6.3. Future Research Directions

6.3.1. Technology Integration Innovation

The deep integration of ETC with emerging technologies represents the most promising development direction in this field. This integration involves not merely simple superposition at the technical level but systematic innovation reconstruction that will bring revolutionary changes to urban traffic management.

ETC + V2X deep integration technology will inaugurate a new era of vehicle–road cooperation. Yang Li noted that expanding ETC applications to vehicle–road cooperation scenarios to provide traffic information services for ETC users will attract more non-ETC users to install ETC electronic tags. This integration can not only expand ETC user bases but more importantly achieve complementary advantages between ETC and intelligent connected vehicle technologies. Mature infrastructure and the substantial user groups of ETC systems provide ready deployment platforms for V2X technology, while the intelligent characteristics of V2X technology provide broader development space for ETC applications.

Smart transportation systems integrating 5G + ETC demonstrate enormous technological potential. The 5G and ETC integration scheme proposed by Zhu Xihao et al. achieves highly intelligent comprehensive traffic management through comprehensive applications of technologies, including big data, cloud computing, modern communication technology, mobile internet technology, and satellite positioning systems. This integration not only enhances data transmission speed and reliability but also provides powerful technical support for applications, including real-time traffic state sensing, road traffic control, urban emergency management, and traffic data analysis [4].

The deep applications of artificial intelligence technology will drive the transformation of ETC data analysis from rule-driven to intelligence-driven approaches. AI technologies, including deep learning, reinforcement learning, and federated learning, demonstrate enormous potential in traffic prediction, state recognition, and decision optimization. These technologies can not only improve analysis precision and efficiency but also address complex problems that are difficult to resolve through traditional methods. However, AI technology applications in ETC data still face challenges, including high data quality requirements, poor model interpretability, and substantial computational resource consumption, requiring in-depth research in algorithm optimization, model compression, and edge deployment.

Blockchain technology applications in ETC data management will provide new solutions for data security and trust establishment. Blockchain characteristics, including decentralization, immutability, and traceability, can effectively address security and trust issues in ETC data multi-party sharing processes. Through constructing blockchain-based ETC data consortium chains, secure data sharing and collaborative analysis can be achieved, providing technical foundations for cross-departmental and cross-regional ETC data applications.

6.3.2. Application Scenario Expansion

Application scenarios for ETC technology in urban traffic management are continuously expanding and deepening, presenting diversified, intelligent, and refined development trends.

ETC applications in intelligent connected vehicle environments will face entirely new technical challenges and development opportunities. Against the background of the gradual proliferation of intelligent connected vehicles, traditional vehicles and intelligent connected vehicles will coexist for extended periods, forming mixed traffic flow environments. ETC systems must adapt to these new traffic environments to provide differentiated services for different vehicle types. This requires not only corresponding adjustments in technical architectures but also innovations in management strategies and service models.

MaaS (Mobility-as-a-Service) platform integration represents an important development direction for ETC applications. By integrating ETC into Mobility-as-a-Service platforms, integrated and personalized travel services can be provided to users. This integration not only simplifies user operational processes but also provides more precise travel recommendations and services through data fusion and intelligent analysis, connecting travel chains, service chains, and management chains. However, MaaS platform construction involves multiple stakeholders and technical systems, requiring the in-depth exploration of data standards, interface specifications, and business models.

MaaS platform integration represents an important development direction for ETC applications. By integrating ETC into Mobility-as-a-Service platforms, integrated and personalized travel services can be provided to users. This integration not only simplifies user operational processes but also provides more precise travel recommendations and services through data fusion and intelligent analysis, connecting travel chains, service chains, and management chains. However, MaaS platform construction involves multiple stakeholders and technical systems, requiring the in-depth exploration of data standards, interface specifications, and business models.

Smart logistics and freight management applications demonstrate the application value of ETC technology in comprehensive transportation systems. Through ETC systems, the precise tracking and management of freight vehicles can be achieved, providing services, including transportation monitoring, route optimization, and cost accounting, for logistics enterprises, while providing management tools, including freight flow monitoring, overload control, and environmental protection control, for government departments. These applications not only improve logistics efficiency but also provide effective technical support for urban freight management.

6.3.3. Theoretical and Methodological Innovation

Facing the unique characteristics of ETC data and complex application requirements, traditional traffic theories and analysis methods require innovative breakthroughs across multiple aspects.

The construction of next-generation traffic flow theories represents fundamental requirements for adapting to ETC data characteristics. Traditional traffic flow theories are primarily based on continuous flow assumptions and macroscopic statistical characteristics, while ETC data provides discrete, identifiable individual vehicle trajectory information. These data characteristic changes require traffic flow theories to transition from macroscopic continuous models to microscopic discrete models and from statistical average descriptions to individual behavior characterizations. New theoretical frameworks must be established to describe traffic flow evolution patterns, interaction mechanisms, and collective behavioral characteristics based on individual trajectories.

Complex network analysis methods possess important value in ETC data applications. Urban transportation systems are essentially complex network systems with characteristics including small-world and scale-free complex network features. Based on ETC data, more accurate traffic network topological structures and dynamic characteristics can be constructed. Through complex network analysis methods, structural characteristics, evolution patterns, and the cascade effects of traffic networks can be studied. These analysis methods not only facilitate a deep understanding of the inherent patterns in urban transportation systems but also provide theoretical guidance for network optimization, resilience enhancement, and risk prevention and control.

Deep reinforcement learning methods provide new technical pathways for traffic optimization control. Traditional traffic control methods are primarily based on rules and experience, making adaptation to complex and variable traffic environments challenging. Deep reinforcement learning can automatically discover optimal control strategies through agent–environment interactive learning. ETC data-based reinforcement learning methods can achieve more intelligent and adaptive traffic control with broad application prospects in signal optimization, route guidance, and demand management.

The recent GNN developments demonstrate the potential for integrating advanced graph-learning techniques with ETC infrastructure. Future research directions include adapting attention mechanisms for ETC gantry networks, developing conjoint learning frameworks for multi-source ETC data fusion, and exploring traffic-aware prediction models that leverage ETC systems’ unique characteristics in urban transportation networks.

7. Conclusions and Discussion

7.1. Summary of Major Research Achievements

This comprehensive review examines the current status and development trends of ETC data-based urban traffic state sensing and analysis technologies. The major achievements of this field are reflected in several key aspects:

(1)

The unique value and technical advantages of ETC data in traffic sensing have been clearly established. ETC systems possess distinctive characteristics, including high vehicle identity recognition accuracy, extensive spatiotemporal coverage, and stable data quality, providing crucial data support foundations for refined urban traffic management. Compared with traditional traffic detection methods, ETC data demonstrates clear advantages in coverage scope, collection precision, and cost-effectiveness.

(2)

A relatively comprehensive technical system framework has been established. Significant progress has been achieved in key technical components, including data processing, feature extraction, state assessment, and multi-source fusion, forming a multi-level sensing and analysis system from microscopic vehicle behavior to macroscopic network operations. The construction of multi-source data fusion frameworks enables effective complementarity between ETC data, floating car data, and video detection data, significantly improving traffic state estimation accuracy.

(3)

The practical effectiveness of typical application scenarios has been validated. Practices in real-time monitoring and signal control optimization, traffic prediction, and artificial intelligence technology applications, and the transformation of highway technology to urban applications demonstrate that ETC-based traffic management can enhance management efficiency and service levels, providing robust support for smart city transportation system construction.

7.2. Recommendations for ETC Industry Development

Based on current technological development status and future trend analysis, the following recommendations are proposed for ETC industry development:

(1)

Policy formulation should strengthen top-level design and comprehensive planning. It is recommended to establish ETC data open sharing policies that promote orderly data circulation and efficient utilization while protecting user privacy. Simultaneously, targeted incentive measures should be introduced to encourage ETC technology promotion and application in urban traffic scenarios, establishing and improving industry standard systems to regulate ETC system construction and operational management.

(2)

Technology development roadmaps should emphasize phased and systematic approaches. Near-term priorities should focus on breakthrough bottlenecks in massive data real-time processing capabilities and enhancing system integration and standardization levels. Medium-term goals should achieve a deep fusion of ETC with emerging technologies, including 5G, artificial intelligence, and vehicle–road cooperation, expanding application scenarios and service capabilities. Long-term objectives should construct ETC-based intelligent transportation systems to achieve the comprehensive intelligent operation of transportation systems.

(3)

Industrial ecosystem construction should form collaborative development patterns. Industrial ecosystem systems featuring collaborative development among equipment manufacturers, system integrators, operational service providers, and application developers, promoting the coordinated development of upstream and downstream industrial chains should be established. Open technology platforms and standardized interfaces to reduce application development barriers and facilitate the rapid emergence of innovative applications should be constructed. Professional talent team cultivation should be strengthened to provide intellectual support for sustainable industrial development.

7.3. Research Prospects

Looking toward the future, ETC data-based urban traffic state sensing and analysis technologies will embrace broader development prospects. The arrival of the ETC 2.0 era marks a comprehensive transformation of this technology from simple toll collection tools to comprehensive traffic management platforms, possessing enhanced sensing capabilities, computing capabilities, and service capabilities.

Against the backdrop of smart city construction, ETC will become crucial infrastructure for urban transportation digital transformation, deeply integrating with other urban management systems to collectively construct urban transportation systems. The deep fusion of emerging technologies, including vehicle–road networking, edge computing, and digital twins with ETC systems, will further promote the intelligent and refined development of urban traffic management.

Under the guidance of sustainable transportation development strategies, ETC-based refined traffic management facilitates improved traffic efficiency and reduced congestion and emissions, representing an important technical pathway for achieving sustainable transportation development. With the advancement of carbon peak and carbon neutrality goals, the application potential of ETC technology in green transportation and low-carbon travel will be further explored and released.

ETC data-based urban traffic state sensing and analysis technologies will achieve greater progress in theoretical innovation, technological breakthroughs, and application expansion, playing increasingly important roles in constructing safe, efficient, green, and intelligent modern urban transportation systems.