Evaluating Traffic Control Parameters: From Efficiency to Sustainable Development

Highlights

What are the main findings?

• A comprehensive framework identifies and evaluates 19 key traffic control parameters, aligned with Sustainable Development Goals (SDGs).
• A three-phase methodology: Expert discussions, Systematic review, and Multidimensional Analysis that was used to validate traffic optimization strategies.

What is the implication of the main finding?

• Traffic parameters can be optimized not only for efficiency but also to promote sustainable urban development and reduce environmental impact
• The study supports the integration of real-time sensor data with adaptive control strategies to improve urban mobility while addressing SDGs.

Abstract

Understanding the interplay between traffic optimization parameters and their alignment with sensors, control algorithms, and Sustainable Development Goals (SDGs) is essential for improving urban traffic management. The appropriate selection of parameters in urban traffic management is crucial to optimize vehicular flow and meet the Sustainable Development Goals (SDGs). To find out which parameters are most commonly used and appropriate, a comprehensive study was conducted, the results of which are presented in this article. This study uses a three-phase approach: qualitative exploration, systematic literature review, and multiple-dimensional analysis. This study’s contributions include a practical five-level framework for traffic optimization addressing congestion problems, the identification of 19 commonly used traffic control parameters, the analysis of their implementations in recent intelligent traffic control systems, and a proposal of trends to orient these parameters towards efficiency and compliance with the SDGs. The results lay the groundwork for creating new parameters or modifying existing parameters so that the parameters are oriented not only towards efficiency in control algorithms or user experience but also towards meeting the SDGs.

1. Introduction

In recent years, there has been an exponential increase in the urban population. Concurrently, technological advancements and increased accessibility have led to a significant rise in the number of vehicles per household. As an example, in the United Kingdom, the growth of vehicles changed from 14% in 1951 to 74% in 2001 [1]. This exponential growth indicates that the number of vehicles will continue to rise in the present and near future.

As a consequence, traffic congestion is becoming more frequent, resulting in slower and more uncomfortable journeys. In response, many cities have intensified efforts to reduce traffic, a major contributor to urban pollution. Furthermore, during the COVID-19 pandemic, there was a significant reduction in pollution levels in major cities. This reduction was directly linked to decreased traffic volumes. For example, during the lockdown in March 2020, Madrid and Barcelona experienced NO₂ concentration reductions of 62% and 50%, respectively, as a result of a drastic 75% decrease in traffic [2]. These findings illustrate the direct impact that traffic reduction can have on urban air quality.

This observation supports the conclusion that reducing the number of vehicles in urban environments can lower pollution levels. In addition to reducing pollution, decreasing the number of vehicles also leads to fewer traffic accidents. However, it is not always feasible to reduce the number of vehicles in a city. Therefore, alternative solutions must be explored to mitigate traffic congestion.

Traffic congestion is a condition in transport networks characterized by slower speeds, longer trip times, and increased vehicular queues [3]. Many measures have been developed to represent the magnitude of traffic congestion on roadways in urban areas. However, there is debate about the most appropriate measure of traffic congestion. The point at which road or intersection capacity equals traffic demand is a critical concept in traffic management, known as “saturation”. Understanding this concept is key for urban planners, transportation engineers, and policymakers to effectively manage traffic congestion.

These algorithms rely on data collected from various sensors that measure traffic conditions. Although there is a wide variety of sensors available, the core function remains consistent: the data obtained from these sensors are utilized to derive specific parameters.

Figure 1 illustrates the process through which sensors collect traffic data to compute the parameters that feed into the control algorithms, as well as measure the efficiency of the implemented control measures.

Traffic control parameters serve as a crucial interface between sensor-acquired data and the decision-making processes of traffic control algorithms. Sensors deployed within the urban environment collect essential information regarding traffic conditions such as vehicle speed, traffic volume, or density. The data collected depend on the characteristics of the sensors deployed. These raw data calculate pertinent parameters such as average waiting time, queue length, or fuel consumption. Traffic control algorithms use these parameters to act dynamically over traffic conditions, for example, adjusting signal timings to optimize traffic behavior, such as vehicle flow. As Figure 1 presents, ultimately, the parameters can contribute towards achieving the Sustainable Development Goals.

Furthermore, the appropriate selection of control parameters is a key aspect of ensuring the efficacy of traffic management algorithms. An incorrect choice can lead to algorithms making suboptimal decisions, resulting in inefficiencies in vehicle flow and increased congestion. Parameters act as the quantitative representation of traffic reality and serve as input on which algorithms base their decisions. Therefore, parameter selection must accurately reflect the aspects of traffic to be optimized, such as average waiting time, queue length, fuel consumption, and emissions. For example, if the queue length parameter is ignored in favor of only average waiting time, an algorithm might prioritize reducing the wait time on some routes while creating excessive queues on others, effectively displacing the problem rather than solving it. Similarly, if sustainability-related parameters, such as emissions or fuel consumption, are not considered, the algorithm’s decisions may not align with SDGs, potentially compromising health and the environment.

In summary, optimizing urban traffic flow is about improving transportation efficiency and contributing to a broader array of sustainable development objectives, making our cities healthier, cleaner, and more resilient. However, it is important to note that implementing these strategies may face challenges such as resistance to change, lack of funding, and complex urban environments, which need to be addressed for successful implementation.

To contextualize the study presented in this article, it is necessary to know the actors involved in the issue of urban traffic management, as shown in Figure 2. Sensors, parameters, and control paradigms depend on the element in which they are located—vehicle, pedestrian, infrastructure, or network. Although all elements need to be addressed globally, urban infrastructure is currently the most sensorized and interconnected component in operation. In this study, we focus on the Infrastructure-to-Infrastructure scope and its interfaces with the different elements. As communication, for example, between vehicles and infrastructure, evolves, it will be necessary to extend studies, such as the one presented here, to more elements.

Vehicle-to-Infrastructure (V2I), Infrastructure-to-Infrastructure (I2I), and Infrastructure-to-Vehicle (I2V) communication systems play a crucial role in urban traffic management. However, the current reliance on I2I communication has certain limitations. This method, which involves traffic lights, sensors, and other infrastructure elements exchanging information to coordinate traffic flow, is effective but lacks real-time responsiveness to dynamic traffic conditions. The system’s centralized control and data exchange between fixed infrastructure points can manage traffic well, but it often needs to adapt more quickly to changing traffic patterns. This is where the potential of Network-to-Infrastructure (N2I) and Network-to-Vehicle (N2V) communication comes in, as they provide a broader connectivity framework that links multiple infrastructure points and vehicles to a central network, ensuring synchronized and optimized traffic flow across larger urban areas.

Soon, the shift towards V2I and I2V communication is expected to enhance traffic management significantly. In V2I scenarios, vehicles will transmit data to the infrastructure, informing traffic management systems of their speed, location, and intended routes. These data will enable more adaptive and efficient traffic control strategies. Additionally, I2V communication will allow infrastructure elements like traffic lights to send information directly to vehicles, such as the remaining green light duration, enabling drivers to adjust their speed accordingly and reduce stop-and-go driving. The integration of Pedestrian-to-Infrastructure (P2I) and Pedestrian-to-Vehicle (P2V) communication is also critical, as it not only enhances the efficiency of pedestrian interactions with vehicles and infrastructure by alerting both to potential conflicts and optimizing traffic light changes to accommodate pedestrian flow, but also significantly improves pedestrian safety.

Looking further ahead, the long-term vision involves Vehicle-to-Vehicle (V2V) communication, where vehicles communicate directly, minimizing the need for infrastructure mediation. In this scenario, infrastructure would mainly support vehicles not yet connected due to cultural or political reasons, leading to a highly autonomous and self-regulating traffic ecosystem. Additionally, Network-to-X (N2X) communication will continue to provide an overarching connectivity framework, ensuring seamless integration and coordination among all elements involved in the urban traffic environment.

The subsequent sections will discuss parameters and their relevance to the Sustainable Development Goals (SDGs). This exploration will highlight the intersectionality between traffic flow optimization parameters and the broader socio-economic and environmental objectives outlined in the SDGs framework, underscoring the potential of traffic management technologies to contribute to sustainable development initiatives.

Following this examination, a comprehensive review of current states and trends will be conducted, encompassing an extensive analysis of parameters and sensing technologies documented in various scholarly articles about traffic flow optimization. This thorough review aims to elucidate the prevailing methodologies and advancements in the field, providing valuable insights into the ongoing research efforts and technological innovations driving the optimization of traffic flow management systems, thereby reassuring the audience about the depth of our research.

Lastly, the discourse will shift towards addressing the challenges and trends anticipated in the future trajectory of these technologies. By forecasting potential avenues for development and innovation, this section will offer critical reflections on the evolving landscape of traffic management technologies, emphasizing the importance of staying engaged and interested in the upcoming trends, thereby fostering a sense of anticipation and curiosity.

This article is structured into several sections to provide a comprehensive analysis of traffic control parameters and their impact on urban sustainability. Section 2 outlines the three-phase methodology, which combines expert discussions, a systematic literature review, and a multidimensional analysis to identify and validate key traffic control parameters aligned with the Sustainable Development Goals (SDGs). Section 3 introduces the decision-making framework, presenting the five levels of urban traffic optimization, from avoiding congestion to resolving congestion. Section 4 describes the abstraction framework, highlighting the relationship between traffic control parameters and the Sustainable Development Goals (SDGs), as well as the key indicators used in this study. Next, Section 5 focuses on the real-world framework, analyzing the sensors and technologies used for traffic measurement. Section 6 discusses the current state in traffic optimization, summarizing recent advancements in the literature and their implications. Finally, Section 7 and Section 8 conclude this study by summarizing the main contributions and outlining potential future research directions.

2. Methodology

This study employed a three-phase methodology to comprehensively address the challenges of urban traffic optimization. This multi-faceted approach ensured this study’s findings were evidence-based and directly relevant to urban traffic optimization and sustainable development challenges. Figure 3 shows the methodology employed to address the findings of this paper. This methodology combines qualitative and quantitative data [4], validating and contextualizing all pertinent information to select the best parameters.

The methodological framework is divided into three main phases for evaluating and optimizing traffic control parameters. This approach provides a robust analysis that bridges the gap between theoretical concepts and practical traffic management. Each phase is described in detail below.

2.1. Phase 1: Qualitative Exploration

This initial phase involved focus group discussions with experts in traffic management and control engineering. The focus groups are periodically held as part of the research project cited in the funding section, PRESECREL: Models and platforms for predictable, secure, and reliable industrial information technology systems. The primary outcomes were as follows:

• Problem Definition and Objectives: The discussions facilitated a clear articulation of the problem of urban traffic congestion and established the research objectives, which included updating traffic measurement parameters in line with SDGs and enhancing the efficiency of traffic control algorithms.
• Problem Characterization: The focus groups also contributed to characterizing the complexities of urban traffic congestion, leading to the development of a conceptual framework, as illustrated in Figure 4. This framework guided the subsequent phases by highlighting key areas for parameter selection.

2.2. Phase 2: Systematic Literature Review

Following the qualitative exploration of the previous phase, a systematic review of the existing literature was conducted [5]. This phase involved two steps: selection of review criteria and a comprehensive search of the references. The selection of literature was guided by specific criteria, ensuring the relevance and quality of the included studies:

• Measurement Viability: Only parameters that could be measured using available traffic detection technologies, including traffic sensors, video cameras, and other relevant devices, were considered.
• Efficiency Impact: Priority was given to parameters that significantly influence traffic efficiency and contribute to congestion reduction, such as average waiting time (AWT), average travel time (ATT), and queue length (QL).
• Efficiency Evaluation: Parameters were selected which could measure the effectiveness of the implemented traffic management strategies.

The comprehensive search involved searching in academic databases using keywords derived from Phase 1, such as “traffic optimization”, “traffic control parameters”, “SDGs”, “sustainable urban mobility”, and “traffic sensors”. This search aimed to identify relevant studies on traffic optimization and the control parameters they employ. These terms were searched in the scientific databases, making a temporal tracing in order to verify that the most recent references were valid and maintained coherence with the original reference.

2.3. Phase 3: Multiple-Dimensional Analysis

The final phase synthesized the insights gained from the previous stages through a multidimensional analysis, leveraging the expertise of the authors.

• Data Integration: The findings from the focus groups and the systematic literature review were integrated, consolidating qualitative and quantitative information. As a result, a list of relevant parameters were obtained. Data integration was laborious, because it involved formula curation, such as checking for redundancy in formulas, looking for the same parameter referred to by different names, and resolving different formulas representing the same parameter.
• Parameter Validation: The parameters identified in the focus groups were cross-validated with the evidence from the literature. The most relevant and viable parameters for measurement were identified, like travel time, fuel consumption, emissions, queue length, and average speed.
• Tabular Representation: This analysis led to the development of tables that illustrate the relationships between the different dimensions of this study, including the selected parameters their measurement methods and their alignment with relevant SDGs which summarizes the relationship between traffic-related SDGs and the parameters that contribute to them.

This section has described the methodology used to address the challenges of urban traffic optimization, with emphasis on the selection of relevant control parameters that are aligned with the SDGs. This methodological approach, which combines qualitative exploration, systematic literature review, and multidimensional analysis, lays the foundation for the following sections.

3. Decision-Making Layer Framework

Urban traffic congestion remains one of the most pressing challenges in modern cities, affecting not only travel time and fuel consumption but also air quality and overall urban mobility. Addressing these challenges requires a structured approach to traffic management that integrates data-driven decision-making and adaptive control strategies.

One of the key strategies for mitigating congestion is traffic signal optimization, which enhances the capacity of streets, intersections, and road networks to handle varying traffic loads over time. Optimizing traffic signal control can mitigate these problems by enhancing the capacity of streets, intersections, and roads to handle varying traffic loads over time. This optimization can be visualized in the Figure 4, where the X-axis represents time, and the Y-axis indicates the capacity of a traffic network to absorb a certain level of traffic load.

The primary objectives of traffic signal optimization are to (1) avoid congestion, (2) predict congestion load, (3) delay the onset of congestion, (4) reduce congestion duration and load, and (5) advance the exit from congestion. By increasing the capacity of the traffic network, the initiation of congestion can be postponed. Consequently, this is accomplished through adaptive signal control methods that dynamically respond to traffic conditions, allowing more vehicles to traverse intersections during peak periods without inducing delays. Optimization techniques are also designed to minimize the duration and intensity of congestion through real-time adjustments to traffic signals based on current traffic data, thus preventing bottlenecks and ensuring a smoother flow of vehicles. However, this approach is most effective in scenarios where the primary road arteries are not already oversaturated. Furthermore, efficient traffic signal management can alleviate congestion once it occurs. One widely used strategy is the coordination of green light timings across multiple intersections, commonly referred to as “green waves”. While this technique can improve overall traffic flow, it is important to consider its impact on intersecting roads. Increasing green light durations at certain intersections inherently reduces red light times on crossing roads, which may lead to queue formations. If these queues grow excessively, they can spill over into adjacent intersections, potentially inducing new congestion patterns. Therefore, the design and implementation of signal optimization strategies must account for these interactions to balance benefits across the entire network. Therefore, an effective optimization of traffic congestion consists of reducing the gray area in Figure 4 as much as possible while ensuring that secondary congestion effects are minimized.

Urban traffic control can be enhanced through five levels: (1) avoiding congestion, (2) predicting congestion, (3) delaying the onset of congestion, (4) mitigating congestion, and (5) resolving congestion. Each level addresses a different aspect of traffic management, utilizing various strategies and technologies to improve overall traffic flow and reduce congestion.

• Avoiding Congestion (Av.): Avoiding congestion involves implementing proactive strategies to prevent traffic congestion from occurring. This can be achieved through effective urban planning, efficient road network design, and intelligent transportation systems (ITSs).
  To manage traffic demand and distribution, techniques such as dynamic route guidance, adaptive traffic signal control, autonomous vehicles, and congestion pricing can be employed to prevent bottlenecks and maintain smooth traffic flow. Several studies have explored different approaches to mitigating congestion.
  Tang et al. [6] analyze transportation planning strategies aimed at preventing congestion in shared spaces, while Karimi et al. [7] propose an optimization-based framework for mitigating traffic congestion through the reconfiguration of road networks in one-way systems. Furthermore, Gai et al. [8] introduce a centralized system for route allocation and guidance for autonomous vehicles, demonstrating its effectiveness in optimizing traffic flow.
• Predicting Congestion (Pr.): Predicting congestion involves using advanced data analytics and forecasting models to anticipate traffic conditions before they become problematic.
  By analyzing historical traffic data, current traffic patterns, and external factors such as weather and special events, traffic management systems can predict when and where congestion is likely to occur. Early prediction is essential to preventing congestion, as it enables traffic controllers to take proactive measures such as adjusting signal timings or providing real-time traffic information to drivers.
  Yin et al. [9] propose a model for short-term traffic trend prediction, allowing for early detection of congestion and optimized travel route planning to mitigate traffic buildup before it occurs.
• Delaying the Onset of Congestion (Del.): Delaying the onset of congestion focuses on extending the period before congestion begins.
  By optimizing traffic signal timings and implementing adaptive signal control methods, the capacity of the traffic network can be increased, allowing more vehicles to pass through intersections during peak times without causing delays. Techniques such as coordinated signal systems and real-time traffic monitoring help manage traffic flow more effectively, thereby delaying the point at which congestion starts.
  Hussain et al. [10] evaluate warning systems designed to reduce start-up lost time at urban traffic signals, contributing to more efficient intersection flow and helping to delay congestion formation.
• Mitigating Congestion (Mi.): Mitigating congestion involves reducing the severity and duration of traffic congestion once it has occurred. This solution can be achieved through real-time adjustments to traffic signals, dynamic lane assignments, and reversible lanes.
  Additionally, providing alternative routes and improving public transportation options can help to distribute traffic more evenly and reduce the overall load in congested areas. Mitigation strategies aim to minimize the impact of congestion on travel times and ensure a smoother flow of vehicles.
  Ben et al. [11] enhance and calibrate a traffic simulation system to model severe congestion in Beijing’s road network, providing valuable insights into mitigation techniques and their effectiveness.
• Resolving Congestion (Re.): Resolving congestion involves implementing measures to completely alleviate traffic congestion in specific situations. Solutions at this level may include emergency response plans, incident management systems, and rapid clearance of accidents and road obstructions.
  Advanced technologies such as connected and autonomous vehicles (CAVs) can also play a crucial role in resolving congestion by improving traffic coordination and reducing human errors.
  Murali et al. [12] propose an intelligent traffic flow control system that utilizes rapid in-vehicle alerts to prevent accidents caused by poor road conditions, contributing to congestion resolution by minimizing disruptions.
  By swiftly addressing the root causes of congestion and restoring normal traffic flow, these measures enhance the efficiency and safety of the urban transportation network.

These five levels of traffic control enhancement provide a comprehensive framework for managing urban traffic more effectively, ensuring that congestion is prevented, predicted, delayed, mitigated, and resolved using advanced technologies and strategic planning.

An effective way to provide a clear and intuitive visualization of the relationships between traffic congestion factors and optimization strategies is through a heatmap, which visually represents the degree of impact each traffic optimization strategy has on different congestion factors. This approach allows for a structured comparison of how well each intervention mitigates specific congestion causes.

To this end, we first identified the key congestion factors that significantly affect urban traffic flow. Based on an extensive literature review and expert validation, the following ten factors were selected as the primary contributors to congestion:

• Excessive demand exceeding road capacity: when traffic volume surpasses infrastructure limits.
• Prolonged travel times: increased journey durations due to congestion during rush hours.
• Inadequate infrastructure: roads and transport systems that fail to meet current traffic demands.
• Inefficient traffic control mechanisms: poor traffic signal coordination and suboptimal intersection management.
• Special events (concerts, sports, and others): temporary but significant traffic surges.
• Accidents caused by human error: crashes leading to sudden congestion buildup.
• Lane reductions (due to roadwork or incidents): decreasing available capacity.
• Weather conditions: adverse meteorological effects on traffic flow.
• System failures: malfunctions in traffic signals, communication networks, or control systems.

These congestion factors were then compared against the five levels of traffic optimization strategies (avoiding, predicting, delaying, mitigating, and resolving) to determine which approaches are most effective for each issue.

To quantify the relative importance of each congestion factor and optimization strategy, we applied the Analytic Hierarchy Process (AHP) [13] through Pairwise Comparison Matrices, using Saaty’s fundamental scale (1–9) [14]. This method allowed us to systematically compare each congestion factor against the others in terms of impact on traffic flow and, separately, each optimization strategy against one another in terms of effectiveness.

The process involved two key comparison matrices. The first matrix shown in

Table 1. Relative weights of congestion factors and optimization strategies.

(a) Congestion Factor Weighting
Criteria	Relative Weight
Demand	26.9%
Rush Hour	20.4%
Infrastructure	13.0%
Traffic Controls	4.1%
Events	8.8%
Accidents	13.4%
Lane Reductions	7.1%
Weather	2.4%
Systems Failures	3.9%
(b) Optimization Strategy Weighting
Level	Relative Weight
Avoiding	39.2%
Predicting	28.2%
Delaying	13.7%
Mitigating	13.7%
Resolving	5.2%

(a) evaluated the congestion factors, determining their relative weight in contributing to urban traffic congestion. Factors such as excessive demand and travel times were given higher weights due to their substantial impact.

The second matrix shown in Table 1(b) assessed the optimization strategies, measuring how effectively each strategy (avoiding, predicting, delaying, mitigating, and resolving) addressed different congestion causes. This matrix allowed us to determine which strategies had the most significant influence on improving traffic flow.

The final priority weights derived from these matrices were used to compute the heatmap values using Equation (1), clearly visualizing the most impactful interventions.

$$H(i,j)=W_C\left(i\right)\times W_S\left(j\right)$$

(1)

where $H(i,j)$ is the heatmap value at row i (congestion cause) and column j (optimization strategy). $W_c\left(i\right)$ represents the normalized weight of congestion factor i, and $W_s\left(j\right)$ represents the normalized weight of optimization strategy j. Finally, the results were normalized to between 0 and 4.

The resulting matrix of impact values was then visualized using a gradient color scheme in

Table 2. Heatmap of traffic congestion causes and the levels of action for optimization. Light blue indicates no relevance, whereas darker shades of blue represent increasing levels of relevance.

Cause	Av.	Pr.	De.	Mi.	Re.
Demand exceeds capacity	4	3	1	1	0
High number of vehicles at rush hour	3	2	1	1	0
Insufficient infraestructure	2	1	1	1	0
Inadequate traffic controls	1	0	0	0	0.
Specific events (concerts, matches)	1	1	0	0	0
Accidents caused by humans	2	1	1	1	0
Lane reductions	1	1	0	0	0
Weather	0	0	0	0	0
System failures	1	0	0	0	0

. Higher values (dark blue) indicate that a strategy is highly effective in mitigating a particular congestion cause. Lower values (light blue) suggest that the strategy has a minimal effect on that cause.

This visualization aids decision-makers in allocating resources efficiently, ensuring that the most impactful strategies are prioritized to improve urban traffic management effectively.

Figure 4 presents five potential objectives for a traffic control algorithm. To determine which of these objectives should be prioritized in the algorithm’s design, the heatmap analysis shown in Table 2 allows for identifying the most influential causes of traffic congestion. Consequently, this helps establish the most relevant parameters for detecting these causes.

The results indicate that the most effective strategy is to prevent congestion whenever possible and, if avoidance is not feasible, to predict it accurately. Thus, the heatmap analysis not only aids in selecting the most appropriate parameters but also provides insights into the devices and sensors that should be employed in a traffic optimization system.

4. Abstraction Layer Framework

Understanding the role of traffic control parameters requires a structured approach that links real-world data with decision-making processes. This section focuses on the conceptual framework that enables the evaluation of these parameters in relation to urban traffic management and the Sustainable Development Goals (SDGs). The first part (Section 4.1) explores the direct relationship between traffic optimization and the SDGs, highlighting how traffic management strategies contribute to broader sustainability objectives. The second part (Section 4.2) provides an overview of the key parameters used in traffic optimization, while the final part (Section 4.3) establishes the connection between these parameters and the SDGs, ensuring that traffic control aligns with sustainable urban development.

4.1. SDGs

This survey aims to consolidate information from a series of papers focused on optimizing traffic flow. It seeks to update the traffic measurement parameters of a city in terms of the Sustainable Development Goals (SDGs) and to study the evolution of these parameters to ensure efficient traffic control algorithm measurement. As previously mentioned, this reduction in traffic flow also leads to a reduction in pollution.

The optimization of urban traffic flow is intricately linked to the United Nations Sustainable Development Goals (SDGs), contributing to various targets beyond the immediate realm of transportation. The impact of traffic optimization on these goals is assessed using the following scale: 1 for Very Low (VL) contribution, 2 for Low (L) contribution, 3 for Medium (M) contribution, 4 for High (H) contribution, and 5 for Very High (VH) contribution, as shown in

Table 3. SDG importance in urban traffic optimization. Those marked with an X are the ones selected for each SDG.

SDG	VL	L	M	H	VH
Goal 1: No poverty	X	-	-	-	-
Goal 2: Zero hunger	X	-	-	-	-
Goal 3: Good health and well-being	-	-	X	-	-
Goal 4: Quality education	X	-	-	-	-
Goal 5: Gender equality	X	-	-	-	-
Goal 6: Clean water and sanitation	X	-	-	-	-
Goal 7: Affordable and clean energy	-	-	X	-	-
Goal 8: Decent work and economic growth	-	X	-	-	-
Goal 9: Industry, innovation, and infrastructure	-	-	-	X	-
Goal 10: Reduced inequality	X	-	-	-	-
Goal 11: Sustainable cities and communities	-	-	-	-	X
Goal 12: Responsible consumption and production	-	-	-	X	-
Goal 13: Climate action	-	-	-	-	X
Goal 14: Life below water	X	-	-	-	-
Goal 15: Life on land	X	-	-	-	-
Goal 16: Peace, justice, and strong institutions	X	-	-	-	-
Goal 17: Partnership for the goals	-	X	-	-	-

. It is important to note that the values in Table 3 reflect the authors’ qualitative assessment based on their expertise and literature review.

Among the most significant contributions of traffic optimization are its impacts on climate action and sustainable urban development.

Goal 13: Climate Action (Contribution Level: VH). Reducing vehicle emissions through improved traffic flow is a crucial climate action. By cutting down on fuel consumption and lowering traffic congestion, these efforts contribute significantly to reducing urban pollution and greenhouse gas emissions.

Goal 11: Sustainable Cities and Communities (Contribution Level: VH). Improving urban traffic management directly supports sustainable cities by enhancing mobility, reducing congestion, and improving overall urban infrastructure. Many cities worldwide are implementing various traffic optimization strategies, such as intelligent transportation systems, congestion pricing, and public transportation improvements, to address growing traffic challenges and create more livable urban spaces.

Beyond these direct contributions, traffic optimization also plays a crucial role in other SDGs:

Goal 9: Industry, Innovation, and Infrastructure (Contribution Level: H). The survey implicitly supports Goal 9 by focusing on innovative traffic infrastructure, such as modular traffic signals. By proposing novel infrastructure solutions and discussing their implementation, this paper aligns with the goal of fostering industry innovation and building resilient infrastructure.

Goal 12: Responsible Consumption and Production (Contribution Level: H). Efficient traffic management promotes responsible consumption by reducing fuel usage and minimizing the environmental impact of transportation. Optimizing traffic flow leads to less wasteful fuel consumption, contributing to more sustainable production and consumption patterns.

Goal 3: Good Health and Well-being (Contribution Level: M). By minimizing driving stress and enhancing traffic flow through reduced red light wait times, consistent speeds, and decreased idle times, traffic optimization contributes significantly to public health and well-being. The reduction in congestion not only alleviates stress associated with heavy traffic, improving mental health for urban residents, but also mitigates air pollution. Vehicles, particularly heavy goods transport, emit higher levels of carbon while idling or in stop-and-go traffic. By optimizing traffic flow and reducing these inefficient driving conditions, emissions can be significantly lowered, leading to better air quality and overall health benefits for urban populations.

Goal 7: Affordable and Clean Energy (Contribution Level: M). Traffic optimization seeks to develop systems that are innovative, efficient, cost-effective, and straightforward to implement. This contributes to the goal of providing affordable and clean energy by promoting energy-efficient travel and reducing unnecessary fuel consumption.

While other SDGs may not be as directly influenced by traffic optimization, the goals above demonstrate the broader impact such measures can have on sustainable development. The SDGs aim to protect the planet, combat poverty, and strive for a more prosperous, just, and sustainable world for future generations. Therefore, it is crucial for policymakers and urban planners to prioritize traffic optimization in their development plans, considering its significant contribution to these goals.

This aligns with contemporary applications like Google Maps, which seek the shortest routes and consider the ecological impact of travel. These applications highlight eco-friendly routes, even if they involve longer distances or times, by prioritizing factors such as fewer stops and smoother traffic flow. This shift from speed to ecological sustainability not only underscores the growing importance of environmental considerations in traffic management but also motivates the audience to prioritize these factors in their own work.

4.2. Main Parameters

Collecting traffic parameters is important for a number of reasons. First and foremost, it allows for the analysis and understanding of traffic patterns and trends, which can inform the development of effective traffic management strategies. For example, if data show that there is a high volume of traffic at a particular intersection during certain times of the day, steps can be taken to improve traffic flow, such as adding turning lanes or traffic signals. Additionally, traffic data can be used to identify bottlenecks and other areas of congestion, which can be addressed through infrastructure improvements or other measures. Traffic data can also be used to improve safety by identifying high-accident areas and implementing measures to reduce the likelihood of accidents. Overall, collecting traffic parameters is crucial for the efficient and safe operation of our transportation systems.

Table 4. Evaluation index acronyms and their contexts.

Acronym	Name	Context
CA	Congestion avoidance	Cross
AWT	Average waiting time reduction	Vehicle
ATT	Average travel time reduction	Vehicle
ADT	Average delay time reduction	Vehicle
QL	Queue length reduction	Street
FC	Fuel consumption	Vehicle
Em	Emissions	Vehicle
Th	Throughput rate	Street
NoV	Number of vehicles	Street
NoS	Number of stops	Vehicle
AS	Average speed	Vehicle
ART	Average response time	Vehicle
ASD	Average stop delay	Vehicle
ARoT	Average reward over time	Cross
Hw	Average headway	Cross
D	Density	Street
EE	Estimation error	Cross
FR	Flow rate	Cross
PR	Penetration rate	Cross

presents the evaluation index identified and utilized in various studies. The specific papers referencing these parameters will be cited in Section 6, Current States.

Congestion avoidance measures are strategies designed to reduce the likelihood of congestion occurring in the first place or to minimize its impacts when it does occur. Several factors can optimize traffic flow and reduce congestion, including the following:

• Intelligent transportation systems (ITSs): ITS technologies such as traffic sensors, cameras, and intelligent traffic signals can help to optimize traffic flow by providing real-time information about traffic conditions, identifying bottlenecks, and adjusting traffic signal timings to improve traffic flow.
• Road pricing: imposing tolls or charges for using certain roads or highways can help to optimize traffic flow by reducing the number of vehicles on the road during peak hours and encouraging the use of public transportation or carpooling.
• Public transportation: encouraging the use of public transportation can help to optimize traffic flow by reducing the number of vehicles on the road, particularly during peak hours.
• Carpooling and other sustainable transportation options: encouraging carpooling and promoting other sustainable transportation options such as biking and walking can help to optimize traffic flow by reducing the number of vehicles on the road.
• Smart traffic management: smart traffic management systems can be used to optimize traffic flow by adjusting the timing of traffic signals, controlling traffic flow, and providing real-time information to drivers. Implementing intelligent transportation systems (ITSs) to improve traffic measurement and estimation.
• Road design: proper road design can optimize traffic flow by separating different types of traffic and providing dedicated lanes for buses and bicycles.
• Flexible working hours: encouraging flexible working hours can help to optimize traffic flow by spreading out peak hour traffic over a longer period of time.
• Telecommuting: promoting remote working or telecommuting can help to optimize traffic flow by reducing the number of vehicles on the road during peak hours.

Overall, these measures can help to optimize traffic flow, reduce congestion, and promote sustainable transportation options.

Congestion avoidance refers to the efforts to reduce the recurrent daily hours of delay caused by traffic congestion in urban areas. In other words, it focuses on minimizing commuters’ time in traffic, leading to more efficient transportation systems. Congestion occurs when traffic demand exceeds the capacity of the road network. Traffic congestion is the travel time or delay in excess of that normally incurred under light or free-flow travel conditions [15]. It results in slower travel speeds, longer travel times, and increased frustration for commuters. Excessive congestion occurs when the marginal costs to society (such as lost productivity, increased pollution, and stress) outweigh the marginal costs of efforts to reduce congestion (such as investing in better infrastructure or implementing traffic management strategies) [16]. Therefore, the congestion index is the percentage of congestion in an area, following Equation (2).

$$CongestionIndex={\displaystyle \frac{{\sum }_{veh}Dela{y}_{veh}}{{\sum }_{veh}TravelTim{e}_{veh}}}$$

(2)

The total delay represents the cumulative time lost due to congestion (e.g., waiting at intersections and slow-moving traffic). The total travel time is the time it takes to travel without congestion. A higher congestion index indicates more severe congestion. An ideal value of 0 means that there was no congestion.

Average waiting time [17] refers to a vehicle’s time waiting in a queue before the traffic signal turns green at an intersection or any other point in the road network, as shown in Equation (3).

$$AWT={\displaystyle \frac{{\sum }_{veh}WaitingTim{e}_{veh}}{NumberOfVehicles}}$$

(3)

AWT impacts overall traffic flow efficiency. Reducing AWT leads to smoother traffic movement, less congestion, and improved travel commuter travel experience. Low AWT values signify the absence of traffic interruptions. A value of 0 indicates the complete inactivity of vehicles within the system, meaning no vehicle has stopped.

Average travel time represents the average time a vehicle travels from one point to another, represented in Equation (4), considering both moving time and waiting time (if any). Efficient transportation systems aim to minimize average travel time by optimizing traffic flow, reducing congestion, and providing efficient public transportation options [18]. Ideally, the aim is for the value of ATT to be equal to Move Time, which represents the minimum time required for a vehicle to travel from one point to another.

$$ATT={\displaystyle \frac{{\sum }_{veh}TravelTim{e}_{veh}}{NumberOfVehicles}}$$

(4)

Average delay time represents the additional time spent due to congestion, bottlenecks, or other factors that slow down travel, as shown in Equation (5). It quantifies the impact of delays on overall system efficiency and user experience [19]. Like ATT, the objective is to minimize the difference between Travel Time and Move Time, which is the time spent in motion, aiming for values close to zero.

$$ADT=\sum _{veh}(TravelTim{e}_{veh}-MoveTim{e}_{veh})$$

(5)

These times are parameters of the same measurement but in different stages. Therefore, both the objectives and their reduction can be obtained from both. These average times can be used to measure the efficiency of a system or process and identify opportunities for improvement. It is often used in the context of customer service, transportation, and healthcare, among other areas. In order to reduce average times, it may be necessary to increase capacity, streamline processes, or implement other efficiency measures.

The average time in traffic can vary significantly depending on several factors, such as the time of day, the location, and the road conditions. In general, however, the average traffic is likely higher during peak hours, when there are more vehicles on the road, and lower during off-peak hours. It may also be higher in areas with heavy traffic congestion, such as large cities, and lower in areas with less traffic. It is important to note that these times can also be affected by other factors, such as accidents or road construction, which can cause delays.

Queue length refers to the total amount of vehicles waiting in a traffic queue, shown in Equation (6), typically at intersections, toll booths, or other points where vehicles come to a stop. It represents the cumulative number of vehicles waiting for their turn to proceed [20].

$$QL=\sum _{veh}StopVehicles$$

(6)

A long queue of vehicles at a toll booth, a potential challenge, may indicate that there are not enough lanes to accommodate the number of vehicles, which could result in delays and increased average waiting time. In order to reduce queue length and improve the efficiency of a system, it may be necessary to increase capacity, streamline processes, or implement other measures, as in the case of average times.

Emission, fuel consumption, and number of stops in traffic are closely related. Emissions refer to releasing pollutants (such as nitrogen oxides, carbon monoxide, particulate matter, etc.) into the atmosphere from vehicles during combustion or other processes [21]. The specific formulas for calculating emissions depend on the type of pollutant (e.g., NOx, CO, PM) and the vehicle type (gasoline, diesel, electric). Generally, emissions are estimated based on vehicle activity (e.g., speed, acceleration, idling time) and emission factors.

Fuel consumption [22] refers to the amount of fuel (gasoline, diesel, etc.) used during operation, as shown in Equation (7). In urban traffic, reducing fuel consumption contributes to energy efficiency and environmental sustainability. The formula for fuel consumption depends on factors such as vehicle type, driving conditions, and fuel efficiency. The lower the FC, the better the system performance. A simplified formula for fuel consumption is as follows:

$$FC={\displaystyle \frac{DistanceTraveled}{FuelEfficiency}}$$

(7)

The number of stops represents how often a vehicle comes to a complete stop during a journey. In urban traffic, minimizing unnecessary stops improves traffic flow and reduces fuel consumption [16].

In traffic, emissions, fuel consumption, and the number of stops tend to be higher than when driving on the open road. This is because vehicles in stop-and-go traffic constantly accelerate and decelerate, which requires more fuel and results in higher emissions. Additionally, idling in traffic can also increase emissions and fuel consumption.

The pollutants that are most commonly associated with transportation emissions include carbon monoxide, nitrogen oxides, particulate matter, and hydrocarbons. These pollutants can negatively impact human health and the environment, contributing to climate change.

Throughput [23] refers to the maximum flow rate of vehicles or people that a transportation system can handle effectively during a specific period, as shown in Equation (8). It represents the capacity of the system to move vehicles or passengers through a given point or segment. Calculating the throughput is like solving a puzzle. The formula you use depends on the specific context, such as a road segment, intersection, or public transit system. However, for a road segment, the formula is as follows:

$$Throughput={\displaystyle \frac{TotalNumberOfVehicles}{TimePeriod}}$$

(8)

Think of heavy traffic as a bottleneck in a water pipe. The more vehicles on the road, the greater the traffic, and the slower the flow of traffic, reducing the throughput. This can lead to congestion, delays, and increased fuel consumption. Additionally, accidents, roadwork, and other events can also reduce throughput.

The number of vehicles refers to the total number of motorized vehicles (cars, trucks, buses, motorcycles, etc.) in a specific area or on a particular road segment, as shown in Equation (9). It is one of the main factors that affect traffic flow and congestion. The formula is straightforward:

$$NoV=\sum Vehicles$$

(9)

When the number of vehicles in traffic is high, it can lead to congestion, delays, and increased fuel consumption. This is because more vehicles compete for space on the road, which can slow down traffic flow. A higher number of vehicles in traffic can also lead to increased emissions and air pollution.

The average traffic speed [24] refers to the speed at which vehicles travel on a particular road or highway at a given time, as shown in Equation (10). It affects travel time, fuel consumption, and emissions. Values close to the street speed limit indicate significant optimization of traffic flow. It measures the efficiency of a road or highway and can be affected by factors such as traffic volume, lane configuration, and road design. The formula for average speed is as follows:

$$AS=\sum _{veh}{\displaystyle \frac{TotalDistanceTravele{d}_{veh}}{TotalTravelTim{e}_{veh}}}$$

(10)

Generally, average speed tends to be lower in heavy-traffic conditions, as the number of vehicles on the road is more significant and the traffic flow is slowed. This can lead to congestion, delays, and increased fuel consumption. Accidents, roadwork, and other events can also reduce average speed.

Average response time refers to the average time it takes for a transportation system (e.g., traffic signal, emergency services) to react to changing conditions. It impacts efficiency and safety [19]. The formula depends on the specific context (e.g., traffic signal control, emergency response) but is typically measured in seconds or fractions of a second.

Average response time in traffic can be affected by several factors, including the type of vehicle, the driver’s behavior, and the traffic conditions. For example, a vehicle with advanced driver assistance systems (ADASs) may have a faster average response time than a vehicle without ADAS. Additionally, a driver who is alert and paying attention to the road will have a faster average response time than a distracted driver.

There are several factors that can influence the number of vehicles in traffic, such as population growth, urbanization, economic development, and government policies. For example, population growth in an urban area can increase the number of vehicles on the road. At the same time, government policies promoting public transportation, carpooling, biking, and walking can reduce the number of vehicles in traffic.

Average stop delay represents the average additional time vehicles spend stopped or moving very slowly due to congestion, traffic signals, or other factors [25]. It directly affects travel time, user experience, and fuel consumption. It is important to note that the lower the values of average stop delay, the smoother the traffic flow. This is the ultimate goal in traffic management, and understanding this concept is a step towards achieving it. The formula, shown in Equation (11), for average stop delay depends on the specific analysis method (e.g., simulation models and field observations). However, in general, it is represented as follows:

$$ASD={\displaystyle \frac{{\sum }_{spee{d}_{veh\simeq 0}}StoppedTim{e}_{veh}}{NoV}}$$

(11)

In general, the longer the average stop delay, the more fuel a vehicle consumes and the more emissions it produces. It also contributes to delays and inconvenience for the driver and can cause a ripple effect on traffic flow, creating congestion downstream.

Factors influencing the average stop delay include the traffic volume, the road layout and design, and the traffic signal timing. For example, a busy intersection with high traffic volume will have longer stop delays than a less busy intersection.

Average reward over time in traffic can refer to the average reward received by an agent making decisions related to traffic flow, such as a traffic management system [26]. In this context, the agent can be a traffic control system or a self-driving car, and the rewards can be based on factors such as fuel efficiency, travel time, and level of traffic congestion. The agent’s decisions can include things like traffic signal timings, routing decisions, and speed control. The formula varies based on the specific context and the reward metrics considered.

The average reward over time in traffic can be an essential metric to track, as it provides insight into how well the agent can optimize traffic flow and improve transportation efficiency. It can be used to evaluate the performance of different traffic management strategies or algorithms.

Average headway [27] in traffic refers to the average time interval between the passage of two consecutive vehicles at a specific point on the road, shown in Equation (12). It is a measure of the spacing of vehicles on a road or highway and is often used to assess traffic flow and congestion. It represents the time gap or spacing between vehicles as they move along a road segment. The average headway can be calculated using the following formula:

$$Hw={\displaystyle \frac{TotalTimeGap}{NoV}}$$

(12)

In general, the lower the average headway, the higher the traffic volume and the greater the likelihood of congestion. Conversely, the higher the average headway, the lower the traffic volume and the lower the likelihood of congestion.

Factors influencing average headway include the traffic volume, the road layout and design, and traffic signals or roundabouts. For example, a road with heavy traffic volume will have a lower average headway than a road with light traffic volume.

Traffic density [28] refers to the number of vehicles occupying a given length of road or highway at a given time, shown in Equation (13). It is a measure of how congested a road or highway is and can be affected by factors such as traffic volume, road layout and design, and weather conditions. The formula for density is straightforward:

$$D={\displaystyle \frac{NoV}{LenghtofRoadway}}$$

(13)

The higher the traffic density, the more likely traffic will be slowed and congested. This can lead to increased travel times, fuel consumption, and emissions.

Factors influencing traffic density include population growth, urbanization, economic development, and government policies. For example, population growth in an urban area can increase the number of vehicles on the road. At the same time, government policies that promote public transportation, carpooling, or biking and walking can reduce traffic density.

Estimation error in traffic refers to the difference between the actual traffic conditions and the estimated traffic conditions [29]. It can occur when estimating traffic flow, traffic volume, travel time, or other traffic-related variables. The formula for estimation error depends on the specific prediction model used (e.g., machine learning, statistical methods).

Several factors, including measurement error, data uncertainty, and model complexity, can cause estimation errors. For example, traffic flow can be challenging to measure accurately, and minor errors in measurements can lead to significant errors in estimates.

Estimation error can have significant impacts on traffic management and control. For example, suppose traffic flow is estimated to be higher than it is. In that case, traffic signal timings may be adjusted to accommodate the higher flow, leading to unnecessary delays for drivers. If travel time is estimated to be shorter than it is, drivers may be tempted to take a specific route, leading to increased congestion on that route.

Flow rate [30] represents the rate vehicles pass a fixed point on a roadway, shown in Equation (14). It measures the traffic volume on a road or highway and is often used to assess traffic flow and congestion. It is typically measured in vehicles per hour (vph) or vehicles per minute (vpm). The formula for the flow rate is straightforward:

$$FR={\displaystyle \frac{NoV}{TimePeriod}}$$

(14)

Flow rate is an essential metric in traffic management and control, as it provides insight into how well the traffic is flowing and how congested a road or highway is. High flow rates indicate heavy traffic, leading to increased travel times, fuel consumption, and emissions. Low flow rates indicate light traffic and less likelihood of congestion.

Penetration rate in traffic [31] refers to the percentage of vehicles on the road that are equipped with a certain technology or feature, shown in Equation (15). It is often used to track the adoption and implementation of new technologies in the transportation sector, such as advanced driver assistance systems (ADASs), connected vehicles, and electric vehicles. The formula for penetration rate is straightforward:

$$PR={\displaystyle \frac{NumberOfEquippedVehicles}{NoV}}\times 100$$

(15)

The penetration rate of a particular technology in traffic can vary depending on factors such as cost, availability, and government policies. For example, the electric vehicle penetration rate is likely higher in countries with generous government incentives and a well-developed charging infrastructure. The penetration rate is an important metric in the transportation sector as it provides insight into how rapidly drivers are adopting new technologies and how quickly they are being implemented in the transportation system.

4.3. Main Parameters and SDGs

The Sustainable Development Goals (SDGs) are a set of 17 goals established by the United Nations to promote sustainable development and improve the quality of life for people worldwide. Some of the SDGs that are particularly relevant to traffic include the following:

SDG 3: Good Health and Well-being aims to ensure healthy lives and promote well-being for all ages. This goal is relevant to traffic because congestion and accidents can negatively impact the health and well-being of people. Efficient traffic management positively affects public health by reducing waiting times and improving emergency response. Parameters relevant to SDG 3:

• Average Waiting Time Reduction: reducing waiting time at intersections and public transport stops contributes to commuters’ well-being by minimizing stress and frustration.
• Average Response Time: faster response times (e.g., emergency services) directly impact health outcomes and well-being.

SDG 7: Affordable and Clean Energy, which aims to ensure access to affordable, reliable, sustainable, and modern energy for all. This goal is relevant to traffic because transportation is a major energy consumer, and reducing fuel consumption and emissions can promote cleaner energy options. Parameters relevant to SDG 7:

• Fuel Consumption: efficient fuel consumption aligns with energy sustainability and reduced environmental impact.
• Emissions: reducing vehicle emissions supports clean energy goals and mitigates climate change.

SDG 8: Decent Work and Economic Growth is closely related to traffic as well, as it aims to promote sustained, inclusive, and sustainable economic growth, full and productive employment, and decent work for all. Efficient traffic flow and reduced delays positively impact productivity and livelihoods. Parameters relevant to SDG 8:

• Average Reward over Time: positive rewards (e.g., reduced travel time) impact economic productivity and well-being.
• Flow Rate: efficient flow rates support economic activity and sustainable infrastructure.

SDG 9: Industry, Innovation, and Infrastructure, which aims to build resilient infrastructure, promote inclusive and sustainable industrialization, and foster innovation. This goal is relevant to traffic because transportation infrastructure, such as roads and highways, is essential for economic growth and development, and promoting sustainable transportation options can foster innovation. Parameters relevant to SDG 9:

• Throughput Rate: efficient throughput supports economic development and sustainable infrastructure.
• Number of Vehicles: managing vehicle numbers impacts infrastructure planning and innovation.
• Estimation Error: accurate data estimation supports informed decision-making.

SDG 11: Sustainable Cities and Communities, which aims to make cities and human settlements inclusive, safe, resilient, and sustainable. This goal is relevant to traffic because congestion and poor transportation options can negatively impact the livability of communities. Urban traffic parameters directly contribute to achieving these objectives. Parameters relevant to SDG 11:

• Congestion Avoidance: reducing congestion enhances urban mobility and quality of life.
• Queue Length Reduction: shorter queues lead to smoother traffic flow and better accessibility.
• Number of Stops: fewer stops improve urban mobility.
• Average Stop Delay: reducing stop delays enhances overall accessibility.
• Average Headway: optimal headway improves public transport efficiency.
• Density: managing density impacts congestion and urban livability.

SDG 12: Responsible Consumption and Production aims to ensure sustainable consumption and production patterns. This goal is relevant to traffic because promoting sustainable transportation options can contribute to sustainable consumption and production patterns. Managing fuel consumption and emissions in urban traffic contributes to this goal. Parameters relevant to SDG 12:

• Fuel Consumption: efficient fuel use aligns with responsible consumption.
• Emissions: reducing emissions supports sustainable production and consumption patterns.

SDG 13: Climate Action. Transportation is a major contributor to greenhouse gas emissions and air pollution, and traffic congestion can exacerbate these issues. Sustainable urban traffic management plays a vital role in reducing greenhouse gas emissions. Parameters relevant to SDG 13:

• Emissions: reducing emissions from vehicles directly addresses climate change mitigation.

To provide a structured overview,

Table 5. Traffic-related SDGs and relevant parameters.

SDG	Relevance to Traffic	Relevant Parameters
3	Traffic congestion and accidents negatively impact health and well-being. Efficient traffic management reduces waiting times and improves emergency response.	AWT, ART
7	Transportation is a major energy consumer. Reducing fuel consumption and emissions promotes cleaner energy.	FC, Em
8	Efficient traffic flow and reduced delays positively impact productivity and economic activity.	ARoT, FR
9	Transportation infrastructure is key for economic growth. Sustainable transportation options foster innovation.	Th, NoV, EE
11	Congestion and poor transportation reduce urban livability. Optimizing traffic improves mobility and accessibility.	CA, QL, NoS, ASD, Hw, D
12	Sustainable transportation reduces unnecessary fuel use and emissions.	FC, Em
13	Transportation is a major contributor to greenhouse gas emissions and pollution. Traffic management mitigates climate change impact.	Em

summarizes the most relevant SDGs for traffic management, highlighting their significance and the key parameters that contribute to their fulfillment.

Overall, the SDGs are closely related to traffic, and addressing traffic-related issues can contribute to achieving several of the SDGs and improve the quality of life of people.

5. Real-World Framework

In urban traffic control, sensor data collection is just as important as optimal control for reducing traffic flow because it provides the essential, real-time information needed to make informed and efficient decisions. Sensors, such as cameras, inductive loop detectors, and infrared sensors, collect data on traffic density, vehicle speed, and movement patterns. This comprehensive data collection allows traffic control systems to dynamically adjust traffic signal timings, manage lane usage, and predict congestion before it occurs. Without accurate and continuous data collection, any attempts to optimize traffic flow would be based on estimates and assumptions, potentially leading to ineffective management and increased congestion and delays. Therefore, the synergy between data collection and intelligent control is crucial for achieving smooth urban mobility and reducing the negative impact of traffic in cities.

The process of traffic measurement involves the meticulous collection of data related to vehicle movement, density, speed, and other parameters. The accuracy of this traffic data is of utmost significance in transportation planning, congestion management, and safety. The acquisition of these data equips urban traffic control systems to adapt and respond efficiently to real-time conditions. As per Gattuso et al. [32], the most prevalent technologies for vehicular traffic monitoring encompass a variety of sensors.

As traffic sensors, video cameras are in a constant state of recording real-time footage of road conditions [33], thereby offering a spatial–temporal perspective of vehicle flow. These cameras, with the aid of video processing algorithms, facilitate vehicle detection, counting, speed estimation, and traffic monitoring. Pneumatic tubes gauge vehicular traffic by sensing the pressure exerted by passing vehicles [34]. Piezoelectric sensors identify changes in weight and vibration when vehicles traverse them [35]. Inductive loop sensors detect the presence of vehicles using electromagnetic fields [36]. Magnetic sensors pick up disturbances in magnetic fields caused by vehicles [37]. Acoustic sensors utilize sounds to detect vehicles and traffic events [38]. Doppler radar gauges the speed of a moving object using the Doppler effect [39]. Ultrasound sensors detect vehicles by emitting ultrasonic waves and measuring the time it takes for the waves to bounce back from the vehicle, which is used in applications like parking assistance, traffic monitoring, and collision avoidance systems [40,41]. The Global Positioning System (GPS) furnishes the precise location and speed of vehicles [42].

Table 6. Parameters used per sensor. Those marked with an X are the ones selected.

	VC	PT	PS	IL	MS	AS	DR	US	GPS
CA	X			X		X		X	X
AWT	X			X		X		X	X
ATT	X	X	X	X	X	X		X	X
ADT	X	X	X	X	X	X	X	X	X
QL	X		X	X	X	X		X	X
FC		X		X					X
Em	X		X	X	X	X			X
Th		X	X	X	X	X		X	X
NoV	X	X	X	X	X	X	X	X	X
NoS	X		X	X	X	X	X		X
AS	X		X	X	X	X	X		X
ART				X		X		X	X
ASD	X			X		X		X	X
ARoT				X				X	X
Hw	X	X	X	X	X			X	X
D			X	X	X	X		X	X
EE	X	X	X	X	X	X		X	X
FR	X			X	X			X	X
PR	X						X	X	X

shows the parameters that can be obtained from each sensor: VC—video camera, PT—pneumatic tubes, PS—piezoelectric sensors, IL—inductive loop sensors, MS—magnetic sensors, AS—acoustic sensors, and DP—Doppler radar.

Vehicle sensors and infrastructure sensors play distinct roles in traffic management. Vehicle sensors, integrated within vehicles, primarily track specific data such as speed, acceleration, and braking using tools like onboard GPS, wheel speed sensors, and accelerometers. These sensors are advantageous due to their portability and adaptability across different vehicles, but they need enhancements, such as with limited coverage and reliance on vehicle compliance. On the other hand, infrastructure sensors are embedded within road systems, such as pavements and traffic signals, to monitor broader aspects of traffic flow, congestion, and road conditions, with common examples including inductive loop sensors, video cameras, and pneumatic tubes. While these offer the benefits of broader coverage and consistent data collection, they come with higher installation costs and maintenance requirements. The integration of both types of sensors is indispensable for achieving accurate and comprehensive traffic monitoring, with vehicle sensors providing individual vehicle data and infrastructure sensors offering insights into overall traffic dynamics.

6. Current States

Recent advancements in traffic management and optimization have led to the development of diverse approaches aimed at reducing congestion, improving travel efficiency, and enhancing overall urban mobility. In the following, we summarize key contributions in the field, highlighting their methodologies, performance improvements, and the parameters used for evaluation. Table 7 provides a structured overview of these studies, categorizing them based on their proposed approach, achieved outcomes, and key performance indicators.

Du et al. [43] present a monitoring and estimation system, utilizing Vehicular Sensor Networks (VSNs) in Shanghai. VSNs collect data for analysis, though they cannot cover all streets simultaneously, initially limiting accuracy. Significantly, this study introduces two novel patrol algorithms, a key contribution to the field, which effectively reduce detection error from 35% to 10%. System effectiveness is assessed by the number of required vehicles and the estimated error comparison.

Addressing traffic congestion, Knorr et al. [44] propose using beacons that periodically analyze traffic flow and alert drivers to sudden jams, encouraging increased following distances. By employing empirical data from loop detectors on a German Autobahn and Vehicle-to-Vehicle communication (VANET), the study found travel time reductions from 22% to 12%. The primary focus is optimizing travel time.

With a focus on real-world application, Dragoi et al. [45] combine infrastructure sensors (RSU) and vehicular sensors (VANETs) to outline a practical traffic control model aimed at alleviating congestion. Remarkably, this model demonstrates a substantial decrease in average travel time by up to 40% compared to static controls and a significant reduction in emissions by 14% to 40%. Effectiveness is evaluated using the VNSim simulator, with key parameters being emissions and fuel consumption.

The use of SUMO for simulation and a 32-bit AVR microcontroller for algorithm execution is detailed in [46]. Real-time data from intersection sensors are used to run four algorithms: SRPT (Shortest Remaining Processing Time), Fair SRPT, MDDF (Minimum Destination Distance First), and MADDF (Minimum Average Destination Distance First). Results indicate that MDDF and MADDF execution times depend on traffic intensity, SRPT remains constant, and Fair SRPT decreases with increasing traffic intensity due to its congestion control behavior. Key parameters include average, minimum, and maximum execution times and average waiting times per vehicle.

Focusing on real-time traffic management, Xiao et al. [47] utilize the TRED (Traffic Random Early Detection) algorithm to reduce congestion. The system employs sensors to monitor vehicle movement, making it adaptable to various urban environments.

Utilizing two RFID sensors, one at the beginning and another at the end of a specific section of road, Atta et al. [48] describe a system that calculates travel time over specific road sections. It significantly reduces congestion through a fuzzy decision-making system that dynamically adjusts traffic light timings. The measured parameter is density.

Integrating intelligent traffic management, Hartanti et al. [49] introduce a MATLAB simulation control system to optimize traffic congestion through fuzzy control. This approach adjusts parameters such as vehicle number, street width, and queue length, measured using infrared sensors.

In conjunction, Saleem et al. [50] introduce a system leveraging Vehicular Network (VN) communication with Roadside Infrastructure (RSU) to collect traffic data and provide services to drivers, enhancing navigation efficiency with a 95% accuracy and a 5% miss rate.

Addressing the boundary feedback control problem of the ARZ (Aw–Rascle–Zhang) traffic model with relaxation terms, Yu et al. [51] introduce two control designs, DORM (Downstream of the Ramp Metering) and UORM (Upstream of the Ramp Metering), to stabilize oscillations in congested traffic regimes. The key parameters measured include traffic density, number of vehicles per unit length, and traffic speed.

The evaluation of a distributed control algorithm, TAPIOCA (Distributed and Adaptive Intersections Control Algorithm), is detailed in [52]. This algorithm determines the sequence and duration of green lights in a multi-intersection intelligent transportation system (ITS). It aims to reduce average waiting time (AWT), prioritize movements, and synchronize successive lights, such as Green Waves. Simulations using SUMO show that TAPIOCA outperforms other dynamic strategies in reducing AWT. Additionally, the article surveys various methods of measuring vehicles via a wireless sensor network deployed at intersections.

A novel approach to dynamic traffic light control at isolated intersections is introduced in [53], combining Wireless Sensor Networks (WSNs) for real-time monitoring with fuzzy logic controllers. Simulation results indicate that this system reduces vehicle waiting times more effectively than other methods.

Zhou et al. [54] present an adaptive traffic light control scheme that adjusts green light sequences at multiple intersections based on real-time traffic data, including traffic volume, waiting time, number of stops, and vehicle density. Collected via WSNs, the data help increase throughput, lower average waiting time, and reduce the number of stops compared to optimal fixed-time control, actuated control, and adaptive control.

A fuzzy control method for traffic lights is explored in [55], with simulations conducted in MATLAB. The results demonstrate improved vehicle delay times compared to traditional timed control methods. Sensors provide data on the number of vehicles in all lanes, with parameters such as average delay time (ADT), AWT, shortest waiting time, and longest waiting time used for comparison.

The controller described in [56] incorporates the Traffic System Communication Algorithm (TSCA) and the Traffic Signals Time Manipulation Algorithm (TSTMA). These algorithms enable adaptive and efficient traffic estimation through dynamic changes in traffic signal sequences and variations in new traffic infrastructure using WSN. The key parameters for comparison include AWT and queue length.

Proposed algorithms that adapt to traffic flow at intersections are examined in [57]. Using the Green Light District Simulator (GLD) and data collected via WSN, simulations of real-life traffic scenarios generate graphs of average waiting time versus cycles. Results show significant reductions in AWT, indicating effective traffic control.

Harahap et al. [58] use the Poisson process and the M/M/1 principle to analyze queue waiting time at traffic lights. MATLAB–Simulink simulations at a traffic light intersection with a single database aim to model and obtain the AWT of each vehicle, automatically determining the duration of red or green lights.

An adaptive traffic light control method for isolated intersections, considering traffic factors like volume and waiting time, is proposed in [59]. Two main algorithms prioritize either the smallest or largest queues. Comparing AWT and queue length against fixed-time methods shows significant reductions with the adaptive approach.

The real-time traffic control algorithm in [33] utilizes computer vision and machine learning (YOLO) with cameras to collect data. Traffic signal phases are optimized based on queue length and vehicle waiting time, aiming to minimize waiting times while maximizing safety.

A traffic management system using fog nodes to collect data is proposed in [60]. The Efficient Dynamic Traffic Light Control algorithm for Multiple (EDTLCM) intersections calculates optimal green light sequences and durations using WSN and neighboring traffic data from fog nodes. The algorithm aims to optimize AWT, fuel consumption, and throughput.

Moreover, Liang et al. [61] introduce a traffic management system employing a deep learning model to control traffic lights at intersections. Simulated in SUMO and using WSN to collect data, the model’s states are represented by two-dimensional vehicle position and speed values, with actions modeled as a Markov decision process. Rewards are based on cumulative waiting time differences between cycles. The system incorporates multiple optimization elements, including dueling networks, target networks, double Q-learning, and prioritized experience replay, reducing AWT by over 20% from the start of training.

The Efficient Dynamic Traffic Control System (EDTCS) introduced by Bharadwaj et al. [62] comprises three components: traffic control unit (TCU), monitor unit (MU), and roadside unit (RSU). Designed to save travel time and prioritize emergency vehicles, the RSU uses RFID readers to identify emergency vehicles and communicate this to the MU. The MU, equipped with sensors such as proximity switches and RFID tags, counts average and emergency vehicles and sends these data to the TCU. The system’s effectiveness is measured primarily by average travel time (ATT).

A traffic control system for self-driving cars is presented in [63], aiming to minimize fuel consumption and travel time by eliminating traffic lights and using constraints to prevent collisions at intersections. The proposed algorithm, simulated in MATLAB, quickly achieves a globally optimal solution for scenarios with up to 8 cars and 16 nodes, with results indicating significant improvements in travel time and fuel consumption.

Wang et al. [64] propose a new car-following (CF) model, MVCM, for mixed traffic flows of human-driven vehicles (HVs) and connected vehicles (CVs) using V2X communication. Simulated in VISSIM, results show that considering optimal speeds of more front vehicles enhances traffic flow stability, with the optimal number being four. The MVCM model outperforms the FVD model in disturbance resistance and matches the stability of the MHOVA model with fewer considered front vehicles. Higher CV rates correlate with increased average speed and reduced travel and delay times, stabilizing when the CV rate approaches 0.6. The model’s positive impact on traffic characteristics, such as average speed and travel time, is confirmed across various traffic volumes and CV rates.

Ma et al. [65] present a method for estimating queue lengths at intersections using travel time data from video-imaging detectors. Establishing a relationship between maximum delay time and queue length per cycle, the method shows higher precision than existing methods, with maximum and average deviations of 39.36% and 12.25%, respectively, over twenty cycles. The parameters used for comparison include ATT and queue length, highlighting the method’s potential applications in traffic management.

An improved deep reinforcement learning model for traffic light control is proposed by [66]. Tested on a large-scale real traffic dataset obtained from surveillance cameras and simulated in SUMO, the model uses cameras for data collection. The evaluation parameters include average travel time (ATT), average delay time (ADT), queue length, and average reward over time. The model outperforms baseline methods, showcasing improvements in the weighted sum of queue length, delay, and waiting time.

A decentralized coordination graph algorithm, MOA3CG (Multi-step return and Off-policy Asynchronous Advantage Actor–Critic Graph), is introduced in [67] for traffic signal control. Utilizing asynchronous multi-agent deep reinforcement learning, the algorithm makes policies based on current traffic states and historical observations. Simulated in SUMO, the experimental results demonstrate that MOA3CG outperforms other algorithms in reducing average delay and travel time while increasing vehicle throughput, effectively mitigating traffic congestion.

Chen et al. [68] propose an improved variable time headway (VTH) spacing strategy for adaptive cruise control (ACC) and cooperative ACC (CACC) systems. Simulations show that this strategy enhances traffic flow efficiency compared to constant time headway strategies, improving metrics such as ATT, ADT, speed, and headway. Numerical simulations are performed on two traffic scenarios to verify the efficiency of the improved strategy. The results show that the improved VTH strategy is suitable and has advantages compared to the constant time headway and VTH strategies. The study also indicates that incorporating ACC/CACC vehicles into mixed traffic enhances stability, road capacity, and congestion alleviation.

A decentralized traffic signal controller using a Nash bargaining game-theoretic framework is detailed in [69,70]. This controller adapts to dynamic traffic demands by optimizing signal timings through cooperative game theory, where each phase is a player aiming for a mutually agreeable outcome. The first paper conducts several simulations on different scenarios. Compared to other controllers, the results show significant reductions in network-wide average travel time, average total delay, stopped delay, and CO₂ emission levels. The second paper compares the proposed control approach to other controllers, showing significant reductions in total delay and vehicle emission levels.

Bhuvaneswari et al. [71] introduce a novel approach for adaptive traffic signal control, simulated in LabView and compared with an existing fixed-time control scheme using data from WSN. The results show that the proposed approach outperforms the existing one, as evaluated by the average delay time (ADT) and the number of vehicles, which were the primary performance metrics used in this comparison.

A hierarchical multi-agent system for traffic signal control is presented by [72]. The system operates on two levels, with individual traffic signals controlled by agents at the physical level and region controllers at the higher level using reinforcement learning and deep neural networks. The agents in the first level employ reinforcement learning to find the best control policy. The agents in the second level use a deep neural network for traffic state prediction, which is used to decide joint actions for the agents in the first level. Simulated in Aimsun2, a widely used traffic simulation software, with deep network implementation in Keras3, a popular deep learning framework, the method effectively reduces average delay time, demonstrating its practical applicability and effectiveness.

Tan et al. [73] propose a route planning scheme for large-scale Unmanned Aircraft System (UAS) operations in urban air space. The system workflow is as follows: first, generate the shortest paths between origination and destination points via heuristic search, and second, schedule the submitted flights to avoid possible conflicts. Utilizing heuristic search and an Evolutionary Algorithm (EA) for scheduling, the approach aims to minimize total flight delays and collision risks. The simulation indicates that increased flight requests result in higher delays, particularly in smaller simulated areas.

An approach for traffic signal control at isolated intersections balancing efficiency and equity is explored in [74]. Using real-time data from connected vehicles, the algorithm minimizes average vehicle delay and limits maximum delay for individual vehicles. The article proposes a central controller that collects the real-time locations of connected vehicles and uses these data to minimize the average vehicle delay while limiting the maximum delay any individual vehicle may experience. Simulations demonstrate significant reductions in long delays and improved equity at intersections.

Addressing variable lane control at urban intersections, Zhou et al. [75] compare traditional Webster timing with a method based on variable lane control and traffic priority. The study focuses on a specific intersection, and the traffic flow is observed during flat-peak and rush-hour periods. The latter proves more effective during peak traffic periods, significantly reducing average delay time and queue length compared to the former, which performs well only during flat-peak periods.

Zhao et al. [76] propose a new integrated modeling framework for a multi-modal traffic state estimation and evaluation of the disruption impact across all modes under various traffic conditions. The results show that a well-balanced mode shift with flexible routing and early announcements of detours so that travelers can plan can significantly benefit all travelers by reducing delay time by 46%. At the same time, a stable route assignment maintains a higher average traffic flow, and the inactive mode-route choice helps relieve density under traffic disruptions.

Improving traffic flow at urban intersections with a fuzzy logic controller is the focus of [77]. The study compared the performance of a fixed-time control system and a fuzzy controller system regarding total delay and average queue length. The study’s results indicated that the fuzzy controller system could significantly reduce total delay and average queue length compared to the fixed-time control system.

A new model for traffic signal transition during evening peak hours is proposed by [78], compared with three classical smooth transition schemes: the immediate transition scheme, the two-cycle transition scheme, and the three-cycle transition scheme. Simulation results show that the proposed model can reduce vehicle delay and queue length compared to the other three schemes. The average reduction ratio of queue length delay was 13.82%, and the average reduction ratio of queue length was 13.65%, indicating that the proposed model performs better than the other three schemes in emergency signal transition applicability.

Hu et al. [79] introduce a new U-turn design with Advance Left Turn (UALT) to address the issues of insufficient intersection spacing and difficulty in the continuous vehicle lane change. UALT provides a dedicated lane for turning vehicles to make a U-turn without traffic interaction. The effectiveness and applicability of UALT were studied through field data investigation, simulation, and analysis with the VISSIM software. Taking a real signal-controlled intersection as an example, the simulation model was built and compared with a conventional intersection and MUIT (multi-lane U-turn intersection). The results showed that UALT could reduce the delay by 73.48% and 41.48%, and queue length by 84.85% and 41.66%, significantly improving the operation efficiency.

Zhao et al. [80] introduce several novel methods for measuring traffic flow using probability theory. The authors propose a technique to estimate total queue length and traffic volume by establishing a single-variable equation for the penetration rate of probe vehicles. Validation tests conducted with simulated and real-world data indicate that the proposed methods are sufficiently accurate for assisting performance measures and traffic signal control at intersections. Importantly, these methods focus on obtaining queue length rather than optimization.

Comert et al. [81] present a new approach for the real-time prediction of traffic queue length at signalized intersections using an Adaptive Signal Control System (ASCS). The study aims to develop short-term prediction models utilizing six variations of Gray systems. A case study with real-world data from five intersections with adaptive traffic signal control on a calibrated roadway network reveals that Gray models with cosine terms perform as well as or better than more complex models like LSTM and NN in predicting average and maximum queue lengths and estimation error. Parameters used for comparison include queue length, estimation error, root mean squared error, and mean absolute error.

A system that integrates video data analysis, data mining, and expert opinion to optimize variable signal timing profiles, thereby reducing congestion at a signalized intersection, is described in [82]. Tested with real-world data from a congested intersection in Dhaka, the system significantly reduces vehicle queue lengths by 40% and increases average traffic speed. The study emphasizes the effectiveness of combining different data sources and expert insights to enhance traffic signal performance.

Vogel et al. [83] propose an adaptive traffic light controller based on fuzzy logic to improve traffic flow at an isolated intersection. The controller uses data from road detectors, such as queue length, arrival flow, and exit flow, to determine the optimal duration for the next signal phase through a set of fuzzy rules. Tested against a fixed signal program in three scenarios with varying traffic demands, the adaptive controller effectively reduces queue lengths and the number of stops. The study employs both simulated and real-world data to validate the results.

The preceding articles are outlined in Table 7, where the initial column, denoted as “Paper”, corresponds to the article number. The subsequent column, labeled “Proposed approach”, details the system introduced in each article. The third column, “Outcome”, delineates the enhancements realized through implementing the respective systems. Lastly, the fourth and final column, titled “Parameters”, specifies the metrics each system employs to assess its performance.

Table 7. Summary of the proposed approach, outcome, and evaluation index of each paper.

Paper	Proposed Approach	Outcome	Evaluation Index
[43]	Monitoring and estimation approach using VSNs. Two new patrol algorithms for real-dataset analysis.	Detection error decreased from 35% to 10%.	CA, NoV, EE
[44]	Beacons to warn drivers of a sudden traffic jam.	With the communication between vehicles, traveling time reduced from 22% to 12%.	CA, ATT
[45]	Joint system of RSU sensors and VANETs to reduce traffic congestion.	ATT reduced by up to 40% and emissions increased from 14% to 40%.	CA, FC, Em
[46]	Four different algorithms for a real-time evaluation based on the shortest remaining processing time using wireless sensor networks.	Accurate measurement of execution times.	CA, AWT
[47]	TRED algorithm to organize traffic in real-time through a network of wireless sensors.	Reduction in congestion.	CA
[48,49]	Fuzzy algorithm to control the timing of a signal.	Traffic light stop times can be dynamically varied to reduce traffic congestion.	CA, D
[50]	Offer service to drivers to see the state of traffic and thus be able to reduce congestion.	The accuracy of data reception was 95%, and the miss rate was 5%.	CA
[51]	Boundary feedback control problem of the ARZ traffic model.	Stabilization of the congested traffic oscillations.	CA, D, AS, NoV
[52]	TAPIOCA control to define green phases and duration.	AWT reduction in comparison with other strategies.	AWT
[53]	Dynamic traffic control with fuzzy logic controllers.	AWT reduction in comparison with other simulations.	AWT
[54]	Adjust sequences of green lights in multiple intersections in real time.	Much higher throughput, lower AWT, and fewer number of stops.	AWT
[55]	MATLAB simulation with fuzzy control on vehicle delay time compared with traditional time control.	Reductions in ADT and AWT.	AWT, ADT
[56]	TSCA and TSTMA algorithms provide the system with adaptative and efficient traffic estimation.	Reductions in AWT and queue length.	AWT, QL, Th
[57]	GLD simulation in real-life traffic scenarios to generate a graph of AWT vs. cycles.	Reduction in AWT.	AWT
[58]	Poisson process simulated in MATLAB to obtain AWT and automatically set the duration of the red and green traffic lights.	Optimize traffic light phases by reducing AWT.	AWT
[59]	Adaptive traffic light control by two algorithms: priority to the largest queue and priority to the shortest queue.	There was a significant reduction in both algorithms with the queue length and AWT.	AWT, QL
[33]	Real-time traffic control algorithm on the traffic flow with computer vision and machine learning (YOLO) using queue length and waiting time.	AWT and queue length reduction.	AWT, QL
[60]	Traffic green light sequence optimization by collecting data from WSN and neighboring traffic data from the distributed fog nodes.	Reductions in AWT and fuel consumption and maximization of throughput.	AWT, FC, Th
[61]	The actions are modeled as a Markov decision process, and the rewards are the cumulative waiting time difference between two cycles. Propose a double dueling deep Q network (3DQN) with prioritized experience replay.	AWT reduced by over 20% from the start of the training.	AWT
[62]	Propose an Efficient Dynamic Traffic Control System (EDTCS) with traffic control unit (TCU) and RSU.	Travel time was saved, and emergency vehicles like ambulances were prioritized.	ATT
[63]	Simulation in MATLAB to minimize the fuel consumption and traveling time of all drivers.	The proposed algorithm achieved a globally optimal solution.	ATT, FC
[64]	Simulation in VISSIM of a CF model with the mixed traffic flow of HVs and CVs.	Results verify the positive impact of higher CV rates on traffic characteristics.	ATT, ADT, AS
[65]	New method to estimate the queue length for individual signal cycles using travel time data collected by video-imaging detectors.	The new method proposed in this paper may provide advantages for future applications.	ATT, QL
[66]	A more effective deep reinforcement learning model for traffic light control.	The last proposed model outperforms the baseline methods.	ATT, QL, ARoT
[67]	Decentralized coordination graph algorithm, referred to as Multi-step return and Off-policy Asynchronous Advantage Actor–Critic Graph (MOA3CG) algorithm.	The proposed algorithm outperforms the other four state-of-the-art algorithms.	ATT, ADT, Th
[68]	An improved variable time headway (VTH) spacing strategy for the adaptive cruise control (ACC) and cooperative ACC (CACC) system.	Demonstrates the suitability and advantages of the improved VTH strategy compared with the constant time headway strategy and the VTH strategy.	ATT, AS, Hw
[69,70]	A novel decentralized flexible phasing scheme, cycle-free, adaptive traffic signal controller using a Nash bargaining game-theoretic framework.	The algorithm was tested on an arterial network producing statistically significant reductions in total delay and vehicle emission reduction.	ATT, ADT, QL, Em, ASD, NoS, FC
[71]	Proposed approach simulated in LabView software comparing fixed-time control scheme.	Outperforms other methods in terms of ADT and number of vehicles.	ADT, NoV
[72]	A hierarchical multi-agent system with two levels controls traffic signals with a deep neural network.	Experimental results show that the proposed method can reduce the average delay time effectively.	ADT
[73]	A route planning scheme for large-scale Unmanned Aircraft System (UAS) or multiple-drone operations in complex urban air space.	Smaller simulated areas would have a higher delay time due to the limited airspace.	ADT
[74]	A real-time connected vehicle-based traffic signal control algorithm at isolated intersections.	Significantly reduces long delays and inequitable treatment of vehicles at an intersection.	ADT
[75]	A variable lane control method based on traffic priority.	Effectively reduces the average delay time and queue length of intersections in peak periods.	ADT, QL
[76]	A new integrated modeling framework for a multi-modal traffic state estimation and evaluation.	Delay time reduction of 46%.	ADT, FR, FR
[77]	Apply the fuzzy controller to reduce the total delay and average queue length in urban intersections.	The delay and average queue length were remarkably optimized with the fuzzy controller system.	ADT, QL
[78]	Propose a model which is compared with three classical smooth transition schemes.	The model proposed has better performance than the other three classical transition schemes in emergency signal transition applicability.	ADT, QL
[79]	U-turn design with Advance Left Turn (UALT) that provides a dedicated lane to advance the turning vehicle out of the intersection.	Delay and queue length are significantly improved.	ADT, QL, NoS
[80]	Proposes a series of novel methods based on probability theory.	The validation results using both simulation data and real-world data show that the methods would be accurate enough to assist in performance measures and traffic signal control at intersections.	QL, PR, D
[84]	An architecture and algorithm of mobile service computing are proposed for traffic state sensing with DNN cameras.	The final queue length is determined according to the weight of the two computing results.	QL
[81]	Develop short-term queue length prediction models for signalized intersections that can be leveraged by adaptive traffic signal control systems using six variations of Gray systems.	A Gray model with a cosine term can produce performance better than or comparable to more complex models.	QL, EE
[82]	Combined video data analysis, data mining, and expert opinion to optimize variable signal timing profiles and reduce congestion at a signalized intersection based on micro-simulations.	Significantly reduced vehicle queue lengths (40%) and increased average traffic speed.	QL, AS
[83]	An adaptive traffic light controller based on fuzzy logic for improving the traffic flow at an isolated intersection.	Compared with a fixed signal program in three scenarios with different traffic demands, the results demonstrate the effectiveness of the developed decision rules.	QL, NoS

Table 7. Summary of the proposed approach, outcome, and evaluation index of each paper.

Paper	Proposed Approach	Outcome	Evaluation Index
[43]	Monitoring and estimation approach using VSNs. Two new patrol algorithms for real-dataset analysis.	Detection error decreased from 35% to 10%.	CA, NoV, EE
[44]	Beacons to warn drivers of a sudden traffic jam.	With the communication between vehicles, traveling time reduced from 22% to 12%.	CA, ATT
[45]	Joint system of RSU sensors and VANETs to reduce traffic congestion.	ATT reduced by up to 40% and emissions increased from 14% to 40%.	CA, FC, Em
[46]	Four different algorithms for a real-time evaluation based on the shortest remaining processing time using wireless sensor networks.	Accurate measurement of execution times.	CA, AWT
[47]	TRED algorithm to organize traffic in real-time through a network of wireless sensors.	Reduction in congestion.	CA
[48,49]	Fuzzy algorithm to control the timing of a signal.	Traffic light stop times can be dynamically varied to reduce traffic congestion.	CA, D
[50]	Offer service to drivers to see the state of traffic and thus be able to reduce congestion.	The accuracy of data reception was 95%, and the miss rate was 5%.	CA
[51]	Boundary feedback control problem of the ARZ traffic model.	Stabilization of the congested traffic oscillations.	CA, D, AS, NoV
[52]	TAPIOCA control to define green phases and duration.	AWT reduction in comparison with other strategies.	AWT
[53]	Dynamic traffic control with fuzzy logic controllers.	AWT reduction in comparison with other simulations.	AWT
[54]	Adjust sequences of green lights in multiple intersections in real time.	Much higher throughput, lower AWT, and fewer number of stops.	AWT
[55]	MATLAB simulation with fuzzy control on vehicle delay time compared with traditional time control.	Reductions in ADT and AWT.	AWT, ADT
[56]	TSCA and TSTMA algorithms provide the system with adaptative and efficient traffic estimation.	Reductions in AWT and queue length.	AWT, QL, Th
[57]	GLD simulation in real-life traffic scenarios to generate a graph of AWT vs. cycles.	Reduction in AWT.	AWT
[58]	Poisson process simulated in MATLAB to obtain AWT and automatically set the duration of the red and green traffic lights.	Optimize traffic light phases by reducing AWT.	AWT
[59]	Adaptive traffic light control by two algorithms: priority to the largest queue and priority to the shortest queue.	There was a significant reduction in both algorithms with the queue length and AWT.	AWT, QL
[33]	Real-time traffic control algorithm on the traffic flow with computer vision and machine learning (YOLO) using queue length and waiting time.	AWT and queue length reduction.	AWT, QL
[60]	Traffic green light sequence optimization by collecting data from WSN and neighboring traffic data from the distributed fog nodes.	Reductions in AWT and fuel consumption and maximization of throughput.	AWT, FC, Th
[61]	The actions are modeled as a Markov decision process, and the rewards are the cumulative waiting time difference between two cycles. Propose a double dueling deep Q network (3DQN) with prioritized experience replay.	AWT reduced by over 20% from the start of the training.	AWT
[62]	Propose an Efficient Dynamic Traffic Control System (EDTCS) with traffic control unit (TCU) and RSU.	Travel time was saved, and emergency vehicles like ambulances were prioritized.	ATT
[63]	Simulation in MATLAB to minimize the fuel consumption and traveling time of all drivers.	The proposed algorithm achieved a globally optimal solution.	ATT, FC
[64]	Simulation in VISSIM of a CF model with the mixed traffic flow of HVs and CVs.	Results verify the positive impact of higher CV rates on traffic characteristics.	ATT, ADT, AS
[65]	New method to estimate the queue length for individual signal cycles using travel time data collected by video-imaging detectors.	The new method proposed in this paper may provide advantages for future applications.	ATT, QL
[66]	A more effective deep reinforcement learning model for traffic light control.	The last proposed model outperforms the baseline methods.	ATT, QL, ARoT
[67]	Decentralized coordination graph algorithm, referred to as Multi-step return and Off-policy Asynchronous Advantage Actor–Critic Graph (MOA3CG) algorithm.	The proposed algorithm outperforms the other four state-of-the-art algorithms.	ATT, ADT, Th
[68]	An improved variable time headway (VTH) spacing strategy for the adaptive cruise control (ACC) and cooperative ACC (CACC) system.	Demonstrates the suitability and advantages of the improved VTH strategy compared with the constant time headway strategy and the VTH strategy.	ATT, AS, Hw
[69,70]	A novel decentralized flexible phasing scheme, cycle-free, adaptive traffic signal controller using a Nash bargaining game-theoretic framework.	The algorithm was tested on an arterial network producing statistically significant reductions in total delay and vehicle emission reduction.	ATT, ADT, QL, Em, ASD, NoS, FC
[71]	Proposed approach simulated in LabView software comparing fixed-time control scheme.	Outperforms other methods in terms of ADT and number of vehicles.	ADT, NoV
[72]	A hierarchical multi-agent system with two levels controls traffic signals with a deep neural network.	Experimental results show that the proposed method can reduce the average delay time effectively.	ADT
[73]	A route planning scheme for large-scale Unmanned Aircraft System (UAS) or multiple-drone operations in complex urban air space.	Smaller simulated areas would have a higher delay time due to the limited airspace.	ADT
[74]	A real-time connected vehicle-based traffic signal control algorithm at isolated intersections.	Significantly reduces long delays and inequitable treatment of vehicles at an intersection.	ADT
[75]	A variable lane control method based on traffic priority.	Effectively reduces the average delay time and queue length of intersections in peak periods.	ADT, QL
[76]	A new integrated modeling framework for a multi-modal traffic state estimation and evaluation.	Delay time reduction of 46%.	ADT, FR, FR
[77]	Apply the fuzzy controller to reduce the total delay and average queue length in urban intersections.	The delay and average queue length were remarkably optimized with the fuzzy controller system.	ADT, QL
[78]	Propose a model which is compared with three classical smooth transition schemes.	The model proposed has better performance than the other three classical transition schemes in emergency signal transition applicability.	ADT, QL
[79]	U-turn design with Advance Left Turn (UALT) that provides a dedicated lane to advance the turning vehicle out of the intersection.	Delay and queue length are significantly improved.	ADT, QL, NoS
[80]	Proposes a series of novel methods based on probability theory.	The validation results using both simulation data and real-world data show that the methods would be accurate enough to assist in performance measures and traffic signal control at intersections.	QL, PR, D
[84]	An architecture and algorithm of mobile service computing are proposed for traffic state sensing with DNN cameras.	The final queue length is determined according to the weight of the two computing results.	QL
[81]	Develop short-term queue length prediction models for signalized intersections that can be leveraged by adaptive traffic signal control systems using six variations of Gray systems.	A Gray model with a cosine term can produce performance better than or comparable to more complex models.	QL, EE
[82]	Combined video data analysis, data mining, and expert opinion to optimize variable signal timing profiles and reduce congestion at a signalized intersection based on micro-simulations.	Significantly reduced vehicle queue lengths (40%) and increased average traffic speed.	QL, AS
[83]	An adaptive traffic light controller based on fuzzy logic for improving the traffic flow at an isolated intersection.	Compared with a fixed signal program in three scenarios with different traffic demands, the results demonstrate the effectiveness of the developed decision rules.	QL, NoS

7. Challenges and Trends

One of the most prominent emerging trends in traffic flow control is the integration of Artificial Intelligence (AI) and machine learning (ML). These technologies are increasingly being adopted because they can process vast amounts of data in real-time and make informed decisions. For instance, AI and ML are used in traffic prediction, where algorithms like Long Short-Term Memory (LSTM) networks are employed to forecast future traffic patterns based on historical data [85]. In a real-world scenario, this could mean predicting heavy traffic on a particular highway during rush hour. Additionally, real-time traffic optimization is achieved through reinforcement learning (RL) techniques [86], which dynamically adjust traffic signals and reroute traffic according to current conditions. This could be seen in action when traffic lights change their timings based on the current traffic volume. AI and ML also play a crucial role in incident analysis by quickly identifying and responding to traffic incidents through data analysis. An example of this could be the detection of a car accident and the subsequent rerouting of traffic to avoid the affected area.

Another trend that is reshaping traffic control is the advancement in communication technologies, particularly Infrastructure-to-Infrastructure (I2I) communications. These technologies are not just game-changers, but they are paving the way for a more efficient and safer traffic system. Currently, I2I communications enable the real-time collection and transmission of data between traffic control nodes, facilitating real-time data exchange [87]. This capability is already improving decision-making processes, reducing congestion through better coordination between vehicles and intelligent traffic signals. But the future is even more promising, as this will evolve into Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) communications. V2I will revolutionize the interaction between vehicles and traffic management systems, while V2V will allow direct communication between vehicles, further optimizing traffic flow and safety by enabling cooperative driving and real-time hazard warnings.

Measuring the effectiveness of traffic control optimization efforts is essential to ensuring safety and efficiency on roads and highways. Some of the challenges in measuring traffic control optimization include data availability, as collecting accurate and comprehensive data on traffic flow, speed, congestion, and accidents is a challenge, especially in urban areas with high traffic volumes. Analyzing these data to identify patterns and trends and evaluate traffic control measures’ effectiveness can be complex and time-consuming. Additionally, measuring the effectiveness of traffic control optimization for different types of transportation, including cars, buses, bicycles, and pedestrians, presents unique challenges. As new technologies such as connected vehicles, autonomous vehicles, and drones become more prevalent, new methods will be needed to measure their impact on traffic flow and safety.

Several trends in measuring traffic control optimization have emerged to address these challenges. Big data and data analytics are used to analyze large traffic datasets and identify patterns and trends. Real-time monitoring using sensors, cameras, and other technologies allows real-time adjustments to traffic control measures [88]. Simulation and modeling are employed to predict traffic flow and evaluate the impact of different traffic control measures. Developing performance metrics such as travel time, delay, congestion, and safety helps evaluate the effectiveness of traffic control optimization. Public–private partnerships are also being formed to develop innovative ways to measure traffic control optimization.

Parameters collected from various sensors, such as cameras, infrared sensors, and LIDAR systems, are integral to feeding AI and ML algorithms [89]. These parameters are not just data, but they are the building blocks for modeling traffic behavior, creating accurate models that represent traffic flow and driver behavior patterns. Furthermore, these data points are crucial in identifying anomalous patterns, detecting unusual or hazardous behaviors in traffic. But the real heroes of this story are the intelligent control nodes, such as smart traffic lights and traffic control units. These nodes benefit greatly from these parameters, as the intelligence embedded in them allows for more efficient traffic management through adaptive responses, such as automatically adjusting traffic signal timings based on real-time traffic conditions. It is important to note that while AI and ML technologies are becoming more prevalent in traffic management, human operators still play a crucial role in overseeing and managing these systems, ensuring their safe and effective operation. This data-driven optimization is not just a concept, but it is a reality that improves overall traffic efficiency and helps in making informed decisions.

Using Kleinberg networks, a type of network model that incorporates both local and global connectivity properties in traffic management introduces unique opportunities to enhance urban traffic efficiency. These networks are named after their creator, Jon Kleinberg, and are designed to mimic the structure of real-world networks, such as social networks or the Internet. Integrating Kleinberg networks [90] into traffic control systems can improve routing efficiency by leveraging their connectivity properties, making traffic navigation more efficient. Additionally, Kleinberg networks offer scalability advantages, efficiently handling extensive traffic data and vehicles. Reinforcement learning (RL) presents significant benefits for traffic flow optimization. RL algorithms continuously learn and adapt to changing traffic conditions, providing ongoing improvements in traffic management. These algorithms can also balance multiple objectives, such as minimizing travel time and reducing emissions, making them highly versatile for traffic control applications. Long Short-Term Memory (LSTM) networks are particularly effective for analyzing time series data, making them ideal for predicting traffic patterns. LSTM networks enhance traffic prediction accuracy by capturing long-term temporal dependencies and combining historical and real-time data for robust predictions.

Incorporating intelligence into traffic control nodes is justified not only by the significant operational efficiency improvements it offers but also by its cost-effectiveness. Intelligent nodes reduce the need for manual intervention, minimizing errors and improving consistency in traffic control. Data-driven automation and optimization through intelligent nodes substantially enhance operational efficiency. Moreover, the long-term cost savings from reduced energy consumption and improved traffic flow can offset the initial investment in these technologies, making them a financially viable solution for traffic management.

8. Conclusions

This study has identified control framework elements, key traffic control parameters influencing traffic flow optimization, their alignment with the Sustainable Development Goals (SDGs), and current states and trends. These findings demonstrate that parameters selected to input and evaluate control algorithms, such as average waiting time, fuel consumption, and emissions, are crucial to assessing the effectiveness of various traffic management strategies. With the parameters reviewed in this study, it is possible to propose new parameters that allow user-centered planning and compliance with Sustainable Development Goals (SDGs). For example, by incorporating a “health impact” parameter, we can ensure that traffic management strategies prioritize the health and well-being of the community, thereby contributing to the SDG on health and well-being.

However, the implementation of these technologies and methodologies faces several limitations. These include resistance to change by local authorities and the population, which can be addressed through effective communication and public engagement strategies. Funding limitations can be overcome through public–private partnerships or innovative financing models. There is also the need for robust infrastructure to support advanced technologies, which can be met through strategic infrastructure planning and investment. Overcoming these barriers is essential to achieve more efficient and sustainable traffic management.

Future research should focus on validating the proposed framework through real-world case studies, incorporating machine learning techniques for adaptive traffic control, and exploring additional sustainability-oriented parameters. Expanding the scope to vehicles, multi-modal transportation systems, and pedestrian flow analysis could further enhance the applicability of the findings. These aspects involve adopting new technologies, training, and awareness of the stakeholders involved to ensure a smooth transition to more innovative and sustainable traffic systems.