Signaling Improvement Model and Efficiency Analysis of the Model

In recent times, the prevalence of automobiles in metropolitan regions has increased markedly. This considerable increase in vehicular traffic has resulted in a notable expansion of traffic density on existing roadways, thereby rendering the current infrastructure increasingly inadequate. Consequently, there is an ongoing need to develop and implement innovative intelligent transportation systems and advanced control algorithms. These endeavors are designed to mitigate the challenges posed by urban traffic networks, thereby enhancing both the safety and comfort of those who utilize these networks [1-3].

Signalized intersections are a primary source of traffic congestion. The implementation of effective management strategies at these intersections has the potential to markedly reduce overall delay times. It is therefore imperative that innovative techniques for the management of traffic signals are developed and evaluated. The present study proposes a novel signaling optimization control model. The efficacy of the model is evaluated through an analysis of delay parameters and level of service (LOS) metrics. The measurement of delay is a common method of evaluating the efficiency of a signaling system [4-9]. The term denotes the aggregate time spent by a vehicle at an intersection prior to its departure. During this interval, vehicles are compelled to wait due to the signaling system, which disrupts the flow of traffic and results in delays. The term “delay” is typically understood to consist of three components. Initially, as an approaching vehicle responds to changes in the traffic signal, it decelerates, resulting in deceleration delay. Secondly, when the traffic signal switches to red, the vehicle must halt entirely, known as stopping delay. Subsequently, upon the signal’s indication green, the vehicle resumes acceleration and departs the intersection, thereby experiencing acceleration delay. The minimization of delays can be achieved through the optimization of traffic signaling and the improvement of traffic flow. It is postulated that the implementation of adaptive traffic signaling systems and efficient control algorithms has the potential to significantly reduce these delays, thereby enhancing overall traffic circulation efficiency [10]. The accurate and reliable calculation of average delay is of critical importance for the effective traffic planning and regulation of traffic, including the management of traffic flow, the optimization of signalization, and the reduction of congestion. Consequently, researchers and traffic engineers are engaged in ongoing efforts to enhance their comprehension of traffic dynamics and to implement efficacious strategies through the utilization of advanced technologies and data analysis [11]. A variety of analytical techniques, models, and technologies are utilized in the assessment of traffic flow patterns and the estimation of average delay times. Such methods include the utilization of traffic cameras, sensors, and data analytics tools. The data necessary for calculating delays is collected through these technologies, thereby enabling precise assessments of traffic conditions. Moreover, mathematical models such as those put forth by Webster and Akçelik are frequently utilized for optimization of traffic flow and calculations of delays [12, 13].

The Webster delay model is a widely utilized mathematical framework for evaluating the performance of traffic signalization systems, and it has been employed in a multitude of studies. The model is instrumental in calculating delay times and average waiting periods at traffic signals, thereby facilitating the optimization of signal timings at complex intersections and under heavy traffic conditions. The Webster model incorporates factors that account for the time required for the vehicles take to traverse an intersection and the volume of traffic present at that location. The model considers traffic density and demand across a series of intersections when determining optimal signal scheduling. By accounting for the intensity of traffic flow and the impact of signal changes, the model facilitates the optimization of signal timings. The average delay per vehicle is determined by a number of factors, including signal duration, capacity, cycle time, saturation flow, and the number of lanes available for traffic [14-17].

The calculation of delay can be achieved through the utilization of field observations. In this method, researchers or traffic engineers quantify delay by conducting physical monitoring of traffic flow at intersections or specific points within an intersection. The time spent by vehicles passing through and waiting at these locations is manually recorded by observers. The reliability of data obtained from field studies of this nature enables a direct analysis of traffic flow and a more accurate determination of delay [18]. An additional pivotal criterion for assessing the efficacy of signalized intersections is the level of service (LOS). The Highway Capacity Manual (HCM) serves as the standard reference for transportation engineering in the United States, encompassing crucial topics such as roadway and intersection capacity. As part of this guide, the LOS for signalized intersections provides a rating system for the assessment of intersection performance [19, 20]. The LOS is employed to ascertain the efficiency with which an intersection processes traffic. The ratings range from A to F, with A indicating optimal performance and F representing the lowest level of service. Each service level is associated with specific metrics, including vehicle wait times, delay durations, and travel times [21, 22].

Furthermore, simulation programs are also employed to evaluate intersection efficiency and conduct optimization studies. SUMO (Simulation of Urban Mobility) is a widely used open-source software for modeling and analyzing urban transportation systems and traffic flows. It is particularly valuable in research and development activities related to intelligent transportation systems, traffic engineering, and transportation planning [23-25]. Given its open-source nature, SUMO is one of the most extensively utilized simulation programs on a global scale. The SUMO program enables the accurate modeling of real-world intersections, facilitating the operation of these intersections using actual traffic data. The simulation environment allows for the determination of parameters such as delay and wait times, thereby providing valuable insights into the performance of intersections. Signaling systems operate using a variety of methodologies. Fixed-time signaling systems are the most widely used worldwide due to their simplicity and cost-effectiveness. However, these systems frequently demonstrate inefficiencies in terms of performance. In a fixed-time signaling system, the phase durations and number of phases are predetermined, thereby enabling the system to allocate the right of way to traffic from different directions in accordance with set intervals. These fixed intervals are repeated in a cyclical manner. As a result of technological advances, adaptive signaling systems have become increasingly common in the field of intersection management. In contrast to fixed-time systems, adaptive signaling systems are capable of dynamically adjusting phase durations and phase numbers based on real-time data. This data may include, but is not limited to, information regarding intersection traffic density, inter-vehicle spacing, average speeds within the intersection, queuing patterns, data from neighboring intersections, and vehicle presence/absence information [26-29]. The capacity to precisely quantify traffic data and the efficacy of control algorithms contingent upon these measurements are pivotal elements in the functionality of adaptive signaling systems. In this study, an adaptive control system is developed through the creation of a signaling improvement model. The adaptive signaling control model is implemented at a real-world intersection, and its performance is subsequently analyzed. The model employs a fusion of loop sensors and Bluetooth sensors, representing a significant innovation in this research field: the inaugural application of Bluetooth and loop sensor integration in adaptive traffic management. The sensors have been installed at the Polatlı Sanayi Intersection with the objective of facilitating the implementation of an improved signaling model, which will subsequently be operated as the intersection's signaling control system. This study makes a notable contribution to the field by developing an adaptive signal control model that incorporates real-time feedback. The system is designed to continuously evaluate the conditions at the intersection based on data obtained from sensors. This data is subsequently utilized to modify signal timings, and the results of these adjustments are incorporated into an iterative process to enhance the timing algorithm continuously. The performance of the developed signaling improvement model is evaluated based on two key performance indicators: delay and level of service (LOS). The delay is calculated using visual detection from a camera installed at the intersection, and LOS values are derived from these delay measurements. Verification studies are conducted to validate the delay values obtained through visual detection by comparing them with results generated using the SUMO simulation program and Webster's theorem.

The development of adaptive traffic signal control has constituted a central area of investigation within the field of intelligent transportation systems for several decades. A plethora of methodologies and technologies have been developed with the objective of optimizing traffic flow at signalized intersections. The primary goal of these systems is to reduce delays and enhance overall traffic efficiency by dynamically modifying signal timings in response to real-time traffic data. A number of notable approaches to adaptive traffic control have been put forth in the literature. For example, optimization-based methodologies, such as Webster's delay model and its derivatives, seek to calculate optimal cycle lengths and green time allocations with the aim of minimizing delays [30]. These models, however, are based on predefined traffic flow patterns and lack the capacity to adapt to rapidly changing conditions. Rule-based systems such as SCOOT (Split Cycle Offset Optimization Technique) and SCATS (Sydney Coordinated Adaptive Traffic System) are frequently employed as adaptive control methodologies for traffic signal management [31]. These systems dynamically adjust signal timings by utilizing real-time traffic flow data collected from sensors and detectors. While these systems are effective in many scenarios, they often require extensive calibration and infrastructure investment, which makes them less suitable for smaller networks or budget-constrained municipalities. The application of artificial intelligence, including techniques such as reinforcement learning and fuzzy logic controllers, offers a novel approach to adaptive traffic control [32, 33]. These approaches are capable of learning and adapting to traffic patterns over time, however, their deployment is often constrained by computational complexity and the necessity for extensive training data.

A common limitation across these systems is the absence of real-time feedback integration. The majority of adaptive traffic control systems determine signal timings based on sensor data and implement them without assessing their immediate impact on traffic conditions. This unidirectional decision-making process frequently results in suboptimal performance in dynamic and unpredictable traffic environments.

Furthermore, a considerable number of adaptive systems rely on single-sensor configurations, such as loop detectors or radar, which may not provide comprehensive traffic data. Factors such as sensor occlusion, adverse weather conditions, or technical malfunctions can impair their performance.

The integration of Bluetooth and loop sensors in traffic management is a relatively recent development. Bluetooth sensors have been employed to estimate travel times by detecting the Media Access Control (MAC) addresses of Bluetooth-enabled devices [34-36]. This approach offers high accuracy in capturing real-time travel time capture, particularly effective in urban environments with high Bluetooth device penetration among pedestrians and vehicles. However, Bluetooth sensors are not sufficient for capturing vehicle counts or other critical traffic parameters. Loop sensors, on the other hand have been a mainstay in traffic management for decades due to their reliability in detecting vehicle presence and counting traffic volumes [37]. Despite their effectiveness, loop sensors lack the capacity to provide information about travel times or delays, which limits their utility in dynamic traffic scenarios.

The proposed signaling improvement model addresses the limitations of prior systems by combining Bluetooth and loop sensors in a unified framework. The dual-sensor fusion enables the model to capture both travel time data (via Bluetooth sensors) and vehicle count data (via loop sensors), thereby providing a comprehensive dataset for real-time traffic analysis. Furthermore, the model incorporates a feedback mechanism that assesses the efficacy of the implemented signal timings and makes adjustments in a dynamic manner. This iterative process ensures continuous optimization and responsiveness to changing traffic conditions, thereby distinguishing setting it apart from existing adaptive traffic control systems.

The study effectively addresses the critical challenge of optimizing traffic signal timings to reduce delays and improve the level of service at signalized intersections. The proposed signaling improvement model, which integrates real-time feedback through the use of Bluetooth and loop sensors, represents a novel and promising approach to adaptive traffic control. This method is distinguished from conventional systems in that it not only responds to real-time traffic data but also incorporates the performance outcomes of previously applied signal timings into subsequent decision-making cycles. This feedback mechanism ensures continuous optimization, thereby enabling the system to adapt dynamically to varying traffic conditions with greater precision than is possible with traditional models.

In comparison to existing adaptive traffic control systems, which frequently depend on solely static sensor data without feedback integration, this approach represents a significant advancement. The majority of existing systems modify signal timings based on pre-established algorithms or real-time inputs from sensors. However, these systems lack a mechanism to assess the direct impact of these adjustments on traffic flow. The proposed model addresses this limitation by measuring the impact of signal timing on actual travel times and utilizing these findings to refine the control algorithms. This iterative feedback process enhances the system's ability to optimize signal timings, thereby facilitating more efficient traffic management and reducing delays.

Extending the aforementioned signaling improvement model to a more expansive urban network indicates the potential for substantial enhancements in fuel efficiency, a reduction in carbon emissions, and a decrease in the time required for drivers to reach their destinations [39, 40]. For example, the reduction in delay times will result in a decrease in fuel consumption and an increase in efficiency, which will contribute to environmental sustainability. Furthermore, optimized traffic flow can mitigate congestion, enhancing the overall travel experience for road users and potentially reducing the incidence of traffic-related accidents. These quantified benefits serve to underscore the broader significance of the proposed system and its potential to set a new standard for adaptive traffic control solutions.

In this study, the visual detection method was employed to ascertain the delay parameters for the Sanayi Intersection under disparate operating conditions, and the level of service was evaluated in accordance with these delay parameters. A significant aspect of this study is that the observations were conducted within the same time period, with a one-day interval between measurements. This approach ensured that comparable traffic characteristics were employed for assessment of delay parameters for both the fixed-time and the signaling improvement models, thereby maintaining consistency in the evaluated conditions. It is observed that the traffic characteristics at signalized intersections tend to remain consistent within the same time zones on weekdays.

The study found that the signaling improvement model outperformed the fixed-time signaling system in terms of efficiency. The model's capacity to adapt signal timings in accordance with real-time traffic conditions results in enhanced performance, as evidenced by reduced delay values. Furthermore, this adaptive approach yields benefits with respect to the level of service. Table 9 provides a detailed account of the improvement value and percentage for the Sanayi Intersection.

Table 9. Sanayi Intersection improvement value and percentage

Signaling System Operation	Delay Time (s/veh)	Improvement Percentage
Fixed-Time Signaling System	68.18	20.24%
Signaling Improvement Model	54.38

3.1 Verification studies with Webster's theorem

The Webster model was initially employed to substantiate the precision of the visual detection methodology utilized in the investigation. The accuracy of the visual detection method employed in the study was evaluated through the implementation of Webster model calculations for the fixed-time signaling system.

To calculate the delays associated with the fixed-time signaling system, data from the Sanayi Intersection, including cycle time, green times, and peak hour traffic density values, were gathered from the intersection control device before the adaptive signaling system was implemented. This data was collected during the peak hour. In order to minimize delay, the Webster model was employed for the computations.

The Webster delay formula was employed to calculate the average delays per vehicle, with the collected data substituted accordingly [41]. To ensure greater accuracy, the calculations were performed separately for each leg. The formula is provided in Eq. (1).

$d=\frac{C(1-\lambda)^2}{2(1-\lambda x)}+\frac{x^2}{2 q(1-x)}-0.65\left(\frac{C}{q^2}\right)^{\frac{1}{3}} x^{2+5 \lambda}$                 (1)

d: Average delay per vehicle (s/veh),

C: Cycle time (s),

q: Peak flow in the relevant phase (veh/h),

λ: Ratio of green time (G) to cycle time (G/C),

x: Saturation ratio (x = (q/λs) (s=saturation flow))

An illustrative calculation is provided for the purpose of elucidating the Q5 traffic flow.

The value of s (saturation flow), which is employed as a multiplier in the formula was calculated by substituting values such as lane information, speed and slope in Eq. (2), without including parameters related to driver behavior [42].

$s=990+288 T L+8.5 S L-26 G$                   (2)

s: Saturation flow (veh/h)

TL: Number of lanes (3 lanes for Q5)

SL: Speed limit (40 km/h for Q5)

G: Slope (%0 for Q5)

In the formula, the C value represents the total cycle time. The program and timings utilized in the fixed-time signaling system are illustrated in Figure 9. The requisite protection time at the intersection is set to 12 seconds.

9.jpg

Figure 9. Sanayi Intersection times for the fixed-time signaling system

Accordingly, the following values are determined for the Q5 traffic flow.

$\begin{gathered}q=727 \text { veh } / \mathrm{h} \\ C=153 \mathrm{~s} \\ G=55 \mathrm{~s} \\ \mathrm{~s}=2194 \mathrm{veh} / \mathrm{h}\end{gathered}$

When the values are substituted into Eq. (3), the average delay per vehicle for Q5 is calculated.

$d=\frac{153\left(1-\left(\frac{55}{153}\right)\right)^2}{2\left(1-\left(\frac{55}{153}\right)\left(\frac{727}{\left(\frac{55}{153}\right) 2194}\right)\right)}+\frac{\left(\frac{727}{\left(\frac{55}{113}\right) 2194}\right)^2}{2 * 727\left(1-\left(\frac{727}{\left(\frac{55}{153}\right) 2194}\right)\right)}-0.65\left(\frac{153}{727^2}\right)^{\frac{1}{3}}\left(\frac{727}{\left(\frac{55}{153}\right) 2194}\right)^{2+5\left(\frac{55}{153}\right)}$                  (3)

$d=46.92 \mathrm{~s} / \mathrm{veh}$

The delay times were calculated by inputting the peak hour data into Webster's delay formula for each individual leg separately, and the results are presented in Table 10.

In the visual detection method, the average delay per vehicle in the fixed-time signaling system was found to be 68.18 s/veh. Verification calculations conducted in accordance with Webster's theorem yielded an average delay per vehicle for the aforementioned intersection of 64.63 s/veh. As can be observed in Table 11, the outcomes of the visual detection method and the Webster model are found to be approximately analogous. This is a notable finding in regard to the precision of the results obtained from the methodology employed in the present study. The discrepancy between the two values is negligible, amounting to a mere 3.55 s/veh. This discrepancy can be attributed to the fact that the visual detection method depends on data collected from real-world conditions. In other words, external factors exert a more direct influence on the values obtained through the visual detection method. It is to be expected that a slight discrepancy will arise between the results of the visual detection method and those of the Webster method, given that external factors such as road deterioration, pedestrian behavior and driver behavior can exert an influence. The delay value per vehicle in the visual detection method is slightly higher due to the influence of external factors such as road deterioration, pedestrian behavior and driver behavior. As the data in the visual detection method are obtained directly through observation in the field, the margin of error is consequently reduced.

Table 10. Webster's formula delay time results for fixed-time signaling system

Leg of Intersection	Traffic Flow	Delay Time (s/veh)
Leg-1 Ankara	Q7	69.11
	Q5	46.92
Leg-2 City Centre	Q9	67.88
Leg-3 Eskişehir	Q8	71.36
	Q6	52.38
Leg-4 Industry	Q10	80.14
Average Delay Time (s/veh)		64.63

Table 11. Delay values determined by visual detection method and Webster method for fixed-time signaling system

Method	Delay Time (s/veh)
Visual Detection Method	68.18
Webster Method	64.63

3.2 Verification studies with SUMO simulation program

The second method, simulation studies, was conducted to verify the visual detection method. In the simulation studies, the open-source program SUMO was selected as the preferred option. The Sanayi Intersection was operated with a fixed-time signaling system and signaling improvement model. The fixed-time signaling times are illustrated in Figure 9. The data was recorded at varying times, corresponding to the parameters of the signaling improvement model. Additionally, vehicle counts were recorded as data from loop sensors in two different operational scenarios. The signal times and vehicle counts were subsequently input into the SUMO simulation program.

The Sanayi Intersection is modeled in the SUMO simulation program with a one-to-one correspondence. All physical aspects of junction are modelled, including lane widths, lane numbers, turn lanes and geometric features. Furthermore, regarding the visibility of parking and driver behavior at the intersection are incorporated as inputs into the modeled intersection.

The Sanayi Intersection, modeled in the SUMO simulation program, is depicted in Figure 10. The modeled intersection was operated in a simulation environment with traffic data on November 07, 2023 and November 08, 2023 between 17.30 – 18.30. The resulting delay data was obtained from the simulation model, which was run using vehicle counts and signal times for both the fixed-time signaling system and the signaling improvement model during the times of day in operation.

10.png

The simulation results of the Sanayi Intersection fixed-time signaling system and signaling improvement model working methods are presented in Table 12.

In the visual detection method, the average delay per vehicle in the fixed-time signaling system is found to be 68.18 s/veh, while in the signaling improvement model, it is determined to be 54.38 s/veh. As a result of simulation verification method, the average delay value per vehicle for the identical intersection is determined to be 63.79 s/veh for the fixed-time signaling system and as 58.12 s/veh for the signaling improvement model. As evidenced by the results, the outputs of visual detection method and the simulation program are highly correlated. These discrepancies can be observed as external effects in the real-world operations, such as variations in pedestrian behavior or driver conduct.

3.3 General evaluation and related studies

In this study, the visual detection method used to estimate vehicle delays at the Sanayi Intersection was validated against two well-established methods: the Webster delay model and the SUMO simulation program. The objective of this validation was to evaluate the dependability of the delay estimates derived from the visual detection method and to guarantee consistency with the outcomes of widely accepted traffic modeling techniques.

The delay times obtained through the visual detection method were compared with the delay values calculated using Webster's formula and the results from the SUMO simulation program. The visual detection method entailed the tracking of vehicles via fisheye cameras during peak hours, with the time taken for vehicles to traverse the intersection being recorded. The Webster model, a widely used analytical tool, was employed to calculate delays based on cycle time, flow rates, and other intersection parameters. The SUMO simulation tool, which is open-source, was employed to model the traffic behavior at the intersection and to simulate the delays for both the fixed-time and proposed signaling systems.

The comparison of the delay values for each method revealed that the estimates were, for the most part, in close agreement, with only minor discrepancies. In the case of fixed-time signaling system, the visual detection method estimated an average delay per vehicle of 68.18 seconds, whereas Webster's model yielded a delay of 64.63 seconds. The delay values derived from the SUMO simulation for the same system were slightly lower, at 63.79 seconds. The discrepancies between the visual detection method and the other two methods can be attributed to the real-world factors captured in the visual detection, such as driver behavior, road conditions, and pedestrian interactions, which are not fully accounted for in the mathematical models or simulations.

Upon implementation of applying the proposed signaling improvement model, the visual detection method yielded an average delay per vehicle of 54.38 seconds, whereas the SUMO simulation estimated the delay at 58.12 seconds. Once more, the outcomes of both methods exhibited a high degree of concordance, with the sole discrepancy observed between the visual detection and SUMO results pertaining to the signaling improvement model. The discrepancy between the two methods can be attributed to the inherent limitations of each approach, including the simplified traffic assumptions inherent to SUMO.

Table 13 presents a summary of the estimated delays from all three methods for both the fixed-time signaling system and the proposed model.

As illustrated in the table, the estimated delays from all three methods are relatively similar, with the visual detection method yielding slightly higher values. This slight overestimation in visual detection is consistent with the reality of field conditions, which may include factors like minor traffic disruptions and variations in driver behavior that are not fully captured in theoretical models. The consistency among the three methods demonstrates the reliability of the evaluation process and validates the accuracy of the visual detection method as a tool for assessing delays at signalized intersections.

In the study, a signaling improvement model is designed and subsequently implemented at the Sanayi Intersection. The efficacy of the signaling improvement model implemented at the Sanayi Intersection is illustrated through a comparative analysis with the fixed-time signaling system. One of the most significant aspects of this study is the temporal framework within which it was conducted. The method was implemented between 17.30-18.30 at Sanayi Intersection. Upon examination of the daily traffic patterns at Sanayi Intersection, it was determined that 17.30-18.30 time period represents one of the highest density hours. In other words, the results obtained in the study are reflective of conditions occurring during a period when traffic volume is at its peak. Given that the signaling improvement model is designed to adapt to changing traffic conditions, it is anticipated that it will demonstrate enhanced performance in other periods throughout the day. Consequently, when the number of vehicles is low, the signaling improvement model will operate signal times with greater efficiency than the fixed-time signaling system. It can thus be posited that the discrepancy between the per-vehicle delay values between fixed-time signaling operation and signaling improvement model operation will increase when the visual detection method in this study is applied to traffic at other times of the day. As a consequence of the aforementioned, the number of steps among the level of service criteria of the intersection will also increase, given that the difference between delay values per vehicle will increase.

In this study, performance measurement is conducted using a visual detection method and the method is validated through the application of the Webster theorem and SUMO simulation. A principal objective of this study is to present a methodology for evaluating the performance of signaling operating systems. This method is not contingent on the specific signaling operating type. This method allows for the examination of differences in the operation of fixed-time, half-actuated, fully-actuated, and adaptive signaling systems. Furthermore, this method can be employed to investigate the impact of alterations in intersection configurations (geometries) on delay per vehicle. Simulations and theorems are based on parameters such as driver and pedestrian behavior, as well as road imperfections. Nevertheless, this method considers the direct perception of parameters.

Up to now, a multitude of signaling systems and algorithms have been developed with the objective of enhancing their performance. The outcomes of these systems have been quantified and articulated through a variety of methodologies. In these signaling systems, a variety of sensors and data collection methods have been employed. In this study, a signal control algorithm was developed that integrates vehicle path optimization with signal control, utilizing advanced communication technology between approaching vehicles and the signal controller. The MATLAB environment was employed in the study of algorithm. The results demonstrate that the proposed optimization algorithm can enhance intersection performance by reducing vehicle travel time delays, increasing efficiency and capacity. Furthermore, it performs more effectively under congested conditions [43]. In a separate study, an adaptive traffic light control system was developed that employs image processing and image matching techniques in signaling control. In the study, five distinct edge detectors were employed, including Sobel, Canny, Roberts, Log and active contour. The Canny edge process was identified as the best for edge detection [44]. In separate study, the green signal time was calculated using a fuzzy logic controller method that incorporated variables such as the number of motorcycles, cars, and queue length. The findings indicated that the traffic light regulation system employing fuzzy logic controllers demonstrated superior performance compared to the conventional system [45]. In a study, an optimization-based traffic signal cycle length model was developed for signalized intersections. The model was recalibrated using delay parameters in accordance with the Webster delay model, with consideration of the specific traffic conditions in China. The performance of the optimization cycle length model was evaluated in comparison with the TRRL model in terms of delay time and queue length. The findings suggest that the optimization traffic signal cycle length model is more effective than the TRRL model [46]. A modular Timed Synchronized Petri Net (TSPN) model was developed for the implementation of a real-time adaptive control strategy at signalized intersections. The studies were conducted using the SUMO simulation program and the MATLAB program. The study's findings indicate that the proposed control strategy effectively reduces the delay and ensures the safe flow of traffic at the intersection [47]. Another study proposes a heuristic adaptive traffic control algorithm that employs demand estimation based on queue length, with the objective of performing more effectively under varying traffic conditions. The algorithm was analyzed using the VISSIM simulation program. The developed algorithm was evaluated based on two key performance indicators: average control delay and average queue length. Significant improvements are observed in these parameters with the heuristic adaptive traffic control algorithm [48]. In another study, a dynamic control strategy of a signalized intersection is proposed. This strategy enables autonomous adjustment of the green signal time in accordance with the queue length for each lane. The proposed strategy was analyzed using the SUMO simulation program. It can thus be concluded that the proposed dynamic control strategy is more efficient [49]. In another study, an adaptive traffic signal control system based on fuzzy logic controller is proposed. This system utilizes the flow rate received from simple sensors as an input is proposed. The studies were conducted using a micro-scale simulation program. The results demonstrated that the proposed adaptive control model exhibited superior efficiency [50]. In a study, signaling control was conducted using Q-learning method. At the intersection where Q-learning was employed, a notable improvement was observed in terms of reduced queue length and waiting time, as well as a decreased standard deviation of queue lengths [51]. In another study, an adaptive control model for signalized intersections was proposed and subsequently implemented in a real-world setting at a physical intersection. The analyses conducted in the SUMO simulation program revealed that the proposed adaptive model had a beneficial impact on the delay parameter [52]. In another study, a sensor fusion methodology was developed to generate data for use in smart intersections. The sensor fusion comprises image-processing-based cameras and loop sensors. The aforementioned sensor structure has proven effective in a number of applications, including the tracking of object movements, the generation of demand profiles, the detection of incidents, and the enumeration of turn movements [53].

In comparison to other studies in the field, this study makes a significant contribution to the scientific community by offering an adaptive method with feedback. Typically, adaptive signaling systems utilize diverse algorithms and optimization techniques to ascertain signal times based on data obtained from traffic sensors. In this study, the results of the determined adaptive signal times are evaluated and provided as input to the subsequent decision-making algorithm. In other words, the impact of the identified signal times on the travel times on the intersection is quantified and utilized as feedback to inform the subsequent signal time determination process, allowing for a continuous evaluation of the results.

A further innovative aspect of this study is the creation of a sensor fusion consisting of Bluetooth and loop sensors. In adaptive signaling systems, sensors such as cameras, loop sensors, lidar, and radar are typically employed. In this study, Bluetooth sensors and loop sensors are employed in conjunction to generate feedback data and traffic data.

In the signaling improvement model, the objective is to implement a design comprising distinct features and components at the Sanayi Intersection. The implementation of this applied model in urban contexts with different intersection geometries and varying traffic patterns will prove invaluable for its further development.