State-of-Art Review of Trafﬁc Light Synchronization for Intelligent Vehicles: Current Status, Challenges, and Emerging Trends

: The effective control and management of trafﬁc at intersections is a challenging issue in the transportation system. Various trafﬁc signal management systems have been developed to improve the real-time trafﬁc ﬂow at junctions, but none of them have resulted in a smooth and continuous trafﬁc ﬂow for dealing with congestion at road intersections. Notwithstanding, the procedure of synchronizing trafﬁc signals at nearby intersections is complicated due to numerous borders. In traditional systems, the direction of movement of vehicles, the variation in automobile trafﬁc over time, accidents, the passing of emergency vehicles, and pedestrian crossings are not considered. Therefore, synchronizing the signals over the speciﬁc route cannot be addressed. This article explores the key role of real-time trafﬁc signal control (TSC) technology in managing congestion at road junctions within smart cities. In addition, this article provides an insightful discussion on several trafﬁc light synchronization research papers to highlight the practicability of networking of trafﬁc signals of an area. It examines the beneﬁts of synchronizing the trafﬁc signals on various busy routes for the smooth ﬂow of trafﬁc at intersections.


Introduction
For ensuring vehicular traffic safety and consistent flow of traffic, traffic signal control and management at road junctions is a challenging problem in the transportation system. Intersections are the points where two or more routes interact with each other. Intersections are where crossroads, pedestrians, vehicles, and bikes change their direction. Intersections act as an obstruction/barrier for smooth traffic flow within the urban areas because of the large number of vehicles traveling from one place to another. This leads to traffic interruption, congestion, and poor control and management of the traffic [1,2]. The intersection delay affects the signal control logic and the travel efficiency of road users [3]. According to a study, over 10% of the traffic delays all over the world are due to the fixed traffic signal delays. It was found that over 295 million traffic hours of delay was observed on major road intersections in the United States [4]. Therefore, precise estimates of time-dependent delays are needed at junctions on metropolitan roadways for traffic control and management.
Intelligent Transportation Systems (ITS) address the complex issue of effective control and management of traffic at junctions. There should be a balance in between safe and effective traffic control at the intersections in order to allow the maximum vehicles to move through while maintaining safety [5]. Nowadays, traffic-light signaling is used to regulate traffic at crucial junctions/crossings by distributing the same green light timings to all routes [6]. The complicated architecture of traffic systems does not coordinate/link the timings of traffic signals with the average daily road traffic, which leads to congestion at the intersections. Various metropolitan cities such as New Delhi, Bangalore, and Mumbai are going through this imbalanced/uneven traffic flow scenario in cities because the majority of working people live in the neighboring areas. The traffic imbalance is generally seen during the peak hours (morning and evening) when people drive from their jobs and residency. Numerous traffic light management systems have been developed to improve the real-time traffic flow at junctions, but none of them have resulted in real-time traffic synchronization or networking. As a result, metropolitan cities require an updated traffic signal control mechanism/technique that updates traffic signal timing and synchronizes the traffic signals at the road intersections on the basis of real-time traffic information.
For large-scale TSC, synchronization provides a number of advantages. Synchronization is not reliant on any network design or set of rules. This is perfect for network traffic management. When the system's most comparable entities synchronize, it is called synchronization. When traffic signal timing and traffic conditions at nearby intersections are comparable or matching, adaptive and smooth traffic signal coordination can be achieved.
This article provides a comprehensive survey of the TSC strategies that have been developed so far. In addition, it provides a detailed analysis of the associated research study and technology advancement status by comparing the existing techniques used for congestion control at road intersections. This article offers an insightful discussion on several traffic light synchronization research papers to highlight the practicability of networking of traffic signals of an area. It emphasizes the gaps in research, open challenges, and provides potential directions for further research in this area. This paper's key features and contributions are as follows: • To explore the key role of real-time traffic signal control technology in managing congestion at road junctions within smart cities.

•
To summarize the benefits and implementation status of traffic light synchronization and directions for future research for networking traffic lights at intersections of roads.
This article's layout is organized in the following manner. In the beginning, we provide a brief introduction about the history and development background of TSC technology and traffic tight synchronization for intelligent vehicles, and then, the technical benefits of the current traffic light systems are discussed. Then, we compare the existing techniques used for congestion control at road intersections with the help of related work. Afterwards, the synchronization of traffic light development status is summarized and directions for future research for networking traffic lights at road intersections are suggested. Then, typical applications of traffic light networking are discussed. Finally, we conclude the whole paper.

Background of Traffic Signal Synchronization for Intelligent Vehicles
The development of traffic signal synchronization for intelligent vehicles is derived from the traffic light control technology. To better understand the synchronization of traffic lights concept, this section outlines the traffic light control technology and analyzes the advantages of synchronization of traffic lights.

Overview of the Traffic Signal Control Technology
The design of an intelligent traffic light control system is an active research topic. Many researchers are working on the design and development of intelligent traffic signal control systems to solve this stressful issue. Traffic lights can be synchronized as part of regional traffic control methods to enhance the system performance at coordinated junctions. New methods and advanced systems based on artificial intelligence, fuzzy logic, swarm intelligence, evolutionary algorithms [7], image processing, neural network, data fusion, and linear programming etc. have been proposed by the researchers to solve this TSC problem. Table A2 shows the categorization of existing literature on techniques used for congestion control at road intersections. Figure 1 presents the year-by-year trend of the papers published in the area between 1990 and 2021. for congestion control at road intersections. Figure 1 presents the year-by-year trend of the papers published in the area between 1990 and 2021. Numerous traffic light management algorithms have been developed so far to improve the real-time traffic flow at junctions, but none of them have resulted in real-time traffic synchronization or networking. The rising trend reflects the increasing popularity in the subject and emphasizes the importance of current research. Figure 2 illustrates the taxonomy of the paper.

Traditional Traffic Control Methods
Many researchers around the world are working in the field of ITS. It has been known for a long time that working on intelligent traffic monitoring and control systems gives Numerous traffic light management algorithms have been developed so far to improve the real-time traffic flow at junctions, but none of them have resulted in real-time traffic synchronization or networking. The rising trend reflects the increasing popularity in the subject and emphasizes the importance of current research. Figure 2 illustrates the taxonomy of the paper. for congestion control at road intersections. Figure 1 presents the year-by-year trend of the papers published in the area between 1990 and 2021. Numerous traffic light management algorithms have been developed so far to improve the real-time traffic flow at junctions, but none of them have resulted in real-time traffic synchronization or networking. The rising trend reflects the increasing popularity in the subject and emphasizes the importance of current research. Figure 2 illustrates the taxonomy of the paper.

Traditional Traffic Control Methods
Many researchers around the world are working in the field of ITS. It has been known for a long time that working on intelligent traffic monitoring and control systems gives

Traditional Traffic Control Methods
Many researchers around the world are working in the field of ITS. It has been known for a long time that working on intelligent traffic monitoring and control systems gives various traffic-responsive management solutions that may reduce junction delays dramatically. SCOOT [8], SCATS [9], OPAC [10], RHODES [11], PRODYN [12], and MOTION [13] are examples of adaptive traffic control systems that were designed and implemented in several cities in recent decades. SCOOT [8] and SCATS [9] are adaptive traffic control strategies. These systems use loop/magnetic detectors for gathering traffic data. However, the limitation in these methods is that they do not provide the vehicle's speed, heading, and position. While SCATS is a fully computer (microcomputer, regional computer, and central computer)-based system, it is very expensive to use. Unlike SCOOT and SCATS, PRODYN [12] is not based on cyclic settings. PRODYN addresses the global traffic problem using a two-level iterative computation structure after decomposing the enormous initial optimization problem into multiple smaller problems answered via Dynamic Programming.
These adaptive traffic systems generate a "green signal" for large vehicle flow by optimizing the network's traffic signal offset values according to the current traffic demand. Both real-time and predicted traffic arrivals are used in these systems to optimize the objective functions. These systems have significant drawbacks in terms of cost and functionality because they use cameras and loop/magnetic detectors to track vehicles. The initial implementation cost of the system is around $30,000/junction having a cost of $28,800 per mile/year [14]. Moreover, these systems need a large communication infrastructure that is capable of supporting a centralized control with a high data rate. As a result, many cities across the world have developed systems with high implementation cost.
The Webster procedure was used for fixed signal control at isolated junctions [15] for operation based on measured flows. In this method, the cycle length of every intersection is calculated using Webster's equation. The highest cycle length will act as the cycle length of the whole system. This modified demand-based strategy is viable for realtime undersaturated traffic conditions. In fact, the Traffic Response Urban Control (TUC) technique may be utilized to eliminate the predetermined static signal control sequence, which may lead to more successful outcomes than the original TUC approach. However, the issue is that this technique is quite complex.

Vision-Based Traffic Control Methods
A video camera was utilized to predict and monitor real-time traffic using a dynamic Bayesian networks technique in [16]. In this computationally light method, the distribution of spatial interest and spatiotemporal interest points is classified using the Gaussian mixture model (GMM), and then, the dynamic Bayesian approach is used.
Feature-based approaches [17] and neural networks [18] are also used to recognize and track vehicles coming toward an intersection. Feature-based methods track the features less sensitive to partial occlusion. This approach was ineffective for real-time TSC due to its low accuracy and large storage requirements. In these feature-based methods, binarization, rule-based logic, cameras, and road conditions are used to identify vehicles, which makes the process complex and tedious.
Indu et al. [19] proposed a simple but expensive adaptive traffic signaling system that uses video camera data to provide equal green light timings and then allocates enough time using a fair weight and optimal weight calculation algorithm.
Vehicle tracking and image segmentation algorithms become ineffective for real-time operations because of the computational complexity and longer execution time. However, some are not able to operate effectively at night or in variable weather conditions. Zheng et al. [20] use neural networks to predict the upcoming traffic volume from a 15-min daylight traffic flow on an expressway. As a result of the high complexity and memory needs, this approach is unsuitable for real-time systems.

Sensor-Based Traffic Control Methods
Sharma et al. [21] presented a method that assigns equal green signal duration and arrange fair lane departure at an intersection using the GPS data of the users.
Wang et al. [22] proposed a data fusion approach that collects the data on the speed, position, and direction of the vehicles approaching junctions with the help of GPS sensors Other strategies such as detectors [23], on-board GPS, and big data technologies [24,25] are also used to monitor the arrival and departure of vehicles at a junction to reduce the congestion in urban areas. For monitoring the local traffic in a region, sensors and traffic servers are also used. Embedded technology [26] records the GPS data of vehicles and sends it to the traffic monitoring system through GSM/GPRS. This method has a very high implementation cost.
Coll et al. [27] introduced an adaptive system based on linear programming that controls the traffic lights and reduces the waiting time at intersections. This method uses the data gathered from the real-time sensors placed at every intersection.
TSC systems based on Wireless Sensor Network (WSN) have also been used at isolated intersections in [28,29]. Rida et al. [28] proposed a time-fragment based control mechanism that splits the time and calculates the percentage of green ratio. Yousef et al. [30] proposed a connected vehicle-based algorithm for isolated junctions that uses vehicle proximity for releasing vehicle platoons.

Connected Vehicle-Based Traffic Control Methods
Another technique to adaptive decision making is the Virtual Traffic Light (VTL) approach [31], which is a Vehicle-to-Vehicle (V2V) traffic control system with enormous potential, since it may increase traffic flow by more than 30% while eliminating the usage of expensive traffic signals. VTL makes use of V2V communication using the Signal Phase and Timing (SPaT) message and Basic Safety Message (BSM) from the Dedicated Short Range Communication (DSRC) radio for traffic management at intersections. The drawback of this approach is that it requires full penetration of DSRC technology in vehicles, which is not possible at the present time. Furthermore, V2V communications in VTL could come across non-LoS conditions, which makes rapid decision making extremely challenging [32].
A connected vehicle (CV) initiative [33,34] was established by the Virginia Department of Transportation (VDOT) to focus on different V2V and Vehicle-to-Infrastructure (V2I) applications. This program intends to lower the cost of infrastructure for roadside guidance signs and traffic lights by using DSRC technology. However, the complete penetration of the DSRC is required for the connected corridor setup.
Tonguz et al. [35] presented a DSRC-based traffic management method for road intersections. Roads with DSRC-equipped automobiles are given precedence under this system. Even if a small number of vehicles have DSRC technology, this strategy can be effective in reducing the average waiting time on each traffic light. When compared to alternative TSC systems that use detectors, sensors, and cameras, this technique provides a cost-effective solution for traffic management in urban areas by simply using DSRC Road-Side Units (RSUs).
Dresner et al. [36] describes autonomous intersection management (AIM) utilizing multiagent systems, in which drivers and junctions are considered as autonomous entities. Intersections employ a completely new reservation-based technique based on a detailed communication protocol in this mechanism.

Learning-Based Traffic Control Methods
Many studies have recommended using reinforcement learning (RL) to regulate traffic lights and reduce traffic congestion [37,38]. Unlike traditional TSC systems, which depend mainly on predefined models, RL may learn immediately from input as shown in Figure 3. Every junction in RL is portrayed as an agent that optimizes its input-based travel time from the surroundings when the action (i.e., the traffic lights) is set [39].

Learning-Based Traffic Control Methods
Many studies have recommended using reinforcement learning (RL) to regulate traffic lights and reduce traffic congestion [37,38]. Unlike traditional TSC systems, which depend mainly on predefined models, RL may learn immediately from input as shown in Figure 3. Every junction in RL is portrayed as an agent that optimizes its input-based travel time from the surroundings when the action (i.e., the traffic lights) is set [39]. Zheng et al. [40] suggested a FRAP model based on phase competitive modeling, which achieves invariance in traffic flow flipping and rotation conditions. In complicated traffic situations and multi-junction configurations, this system discovers better solutions and delivers greater performance than prior learning methods.
To overcome the fixed traffic signal problem, ref. [41] proposes a novel MARL (Multiagent Reinforcement Learning) approach called Co-DQL, which stands for co-operative double Q-learning. The overestimation problem that affects traditional independent Qlearning is avoided with Co-DQL. TSC simulators are used to test various traffic flow conditions. MARL provides a new reward allocation mechanism for boosting agent stability and resilience.
Joo et al. [42], WeiHua et al. [43], and Zhang et al. [44] investigated some recent improvements in RL techniques that may be applied to solve difficulties with traffic signal regulation. They describe the usage of PPO, A2C, and ACKTR algorithms to handle the problem of partial vehicle detection.
For TSC, Zhang et al. [45] proposes a deep Q-learning approach with partial vehicle identification using DSRC. This system has the benefit of being able to identify cars utilizing different detection technologies such as LTE/5G, BLE 5.0, and RFID.

Miscellaneous Traffic Control Methods
Dezani et al. [46,47] used genetic algorithm (GA) to optimize traffic signals in real time and determine the optimal paths for cars. With the use of a basic neighborhood algorithm, Sabar et al. [48] enhanced GA by using a local search strategy that lowers the average delay in traffic even more than feasible.
In a nine-junction network, Li and Sun [49] suggested a multifunctional GA for traffic optimization. However, this method did not use real-time traffic data.
For traffic light scheduling, refs. [50-52] employ a particle swarm optimization (PSO) technique to regulate traffic lights to minimize the waiting time at intersections so that vehicles can reach their destination in the minimum feasible time. Other swarm intelligence algorithms used for TSC include Ant Colony Optimization [53,54] and Harmony search [55,56].
Milanés et al. [57] presented a V2V communication-based method for reducing traffic congestion at intersections by utilizing the speed and location of autonomous vehicles. Zheng et al. [40] suggested a FRAP model based on phase competitive modeling, which achieves invariance in traffic flow flipping and rotation conditions. In complicated traffic situations and multi-junction configurations, this system discovers better solutions and delivers greater performance than prior learning methods.
To overcome the fixed traffic signal problem, ref. [41] proposes a novel MARL (Multiagent Reinforcement Learning) approach called Co-DQL, which stands for co-operative double Q-learning. The overestimation problem that affects traditional independent Qlearning is avoided with Co-DQL. TSC simulators are used to test various traffic flow conditions. MARL provides a new reward allocation mechanism for boosting agent stability and resilience.
Joo et al. [42], WeiHua et al. [43], and Zhang et al. [44] investigated some recent improvements in RL techniques that may be applied to solve difficulties with traffic signal regulation. They describe the usage of PPO, A2C, and ACKTR algorithms to handle the problem of partial vehicle detection.
For TSC, Zhang et al. [45] proposes a deep Q-learning approach with partial vehicle identification using DSRC. This system has the benefit of being able to identify cars utilizing different detection technologies such as LTE/5G, BLE 5.0, and RFID.

Miscellaneous Traffic Control Methods
Dezani et al. [46,47] used genetic algorithm (GA) to optimize traffic signals in real time and determine the optimal paths for cars. With the use of a basic neighborhood algorithm, Sabar et al. [48] enhanced GA by using a local search strategy that lowers the average delay in traffic even more than feasible.
In a nine-junction network, Li and Sun [49] suggested a multifunctional GA for traffic optimization. However, this method did not use real-time traffic data.
For traffic light scheduling, refs. [50-52] employ a particle swarm optimization (PSO) technique to regulate traffic lights to minimize the waiting time at intersections so that vehicles can reach their destination in the minimum feasible time. Other swarm intelligence algorithms used for TSC include Ant Colony Optimization [53,54] and Harmony search [55,56].
Milanés et al. [57] presented a V2V communication-based method for reducing traffic congestion at intersections by utilizing the speed and location of autonomous vehicles.
Bi et al. [58,59] presented a type-2 FLC system to solve the coordination and complicated uncertainty issues in heavy but organized traffic. (based on fuzzy and Q learning). Firdous et al. [60,61] developed a real-time control system based on fuzzy logic to reduce vehicle waiting times at intersections.

Advantages of Synchronizing the Traffic Lights
The primary goals of TSC technology is to improve traffic safety and travel quality as well as lower the pollution caused by vehicles. However, due to the rapid growth of numerous issues such as the increase in the number of vehicles on the road, traffic jams, Traffic congestion has been a serious concern in metropolitan areas as vehicular traffic has increased. The problem is even worse at road intersections, where pre-timed traffic signals produce delays, increased vehicle activity costs, and air pollution. Time delays are least caused in roadways and rural areas; they are frequently observed in metropolitan cities. Consequently, metropolitan traffic shapes the concerned area of research on delays caused for vehicles and pedestrians. The basic objective of signalized junctions is to decrease the problems that are experienced more at unsignalized crossing points and rotaries and furthermore to give an equivalent chance to all vehicles to cross the intersection.
To address these concerns, traffic light synchronization is a technique in which a vehicle traveling along one side of the road at a specific speed can continue to the opposite end of the road indefinitely by obtaining the maximum number of green lights at the intersections.

Development Status of Synchronization of Traffic Lights
The synchronization of traffic signals across a city is supported by traffic light control frameworks. When automobiles pass through a junction network, these frameworks lower the amount of time they have to wait. Despite the previous research in this domain over the last five decades, improved traffic signal control framework administration is still required.
To enhance traffic flow efficiency at high-density junctions, Hu et al. [62] employed a long green and long red (LGLR) technique for synchronizing the traffic lights of an area. For smoothening the traffic flow across the city, ref. [63] employed a real-time system for the synchronization of traffic lights using multiagent fuzzy logic and the Q-learning approach.
Amogh et al. [64] presented a method for improving traffic synchronization efficiency by dynamically modifying timing settings. This system uses traffic cameras to obtain vehicle density at each intersection and then uses OpenCV to conduct operations on the pictures.
Khan et al. [65] proposes a self-organized TSC system based on sensors. It improves the efficiency of traffic flow and reduces the waiting time at intersections. In comparison to traditional traffic management systems, this technology gives more information and prioritizes the road with a large number of cars at an intersection first. However, this approach does not synchronize the traffic lights of all the intersections.
V2I connectivity was utilized in [66] for the networking of junctions and other road signs. It is used to transmit information on infrastructure such as speed restrictions, road alterations, and traffic light information to smart cars.
Sirikham et al. [67] presented a traffic light synchronization system that used two control boards, a sender and receiver. These control boards use the NodeMCU V.2 (2.4 GHz WiFi) to synchronize the control signals by ESP-NOW protocol, as shown in Figure 4. However, this system works only for single-lane traffic control.
Kumar et al. [68] proposed a synchronization method for two traffic signals using the Long Range (LoRa) module. It uses the time division algorithm to get the traffic information so that the rerouting of vehicles can be done during the peak hours. This concept helps the vehicles reach their destination in the shortest possible time by taking the fastest route.
A synchronization scheme for traffic light controllers based on an explicit finite state model (FSM) is presented in [69]. FSM is modeled using Verilog HDL, and the coding is done using Xilinx Spartan-6 XC6LX16-CS324 FPGA. The machine is merely simulated with six states, which are selected using the traffic control method. Each state provides the appropriate delay, and the requisite traffic signals are turned ON/OFF for that delay. For example, just two routes are picked, and a control algorithm controls the traffic lights on those roads. This study introduced a flexible architecture that uses a clock divider to give a delay in a certain state and discusses the difficulty of modeling the FSM. The topic of selecting a state-encoding technique was also discussed.  Kumar et al. [68] proposed a synchronization method for two traffic signals using the Long Range (LoRa) module. It uses the time division algorithm to get the traffic information so that the rerouting of vehicles can be done during the peak hours. This concept helps the vehicles reach their destination in the shortest possible time by taking the fastest route.
A synchronization scheme for traffic light controllers based on an explicit finite state model (FSM) is presented in [69]. FSM is modeled using Verilog HDL, and the coding is done using Xilinx Spartan-6 XC6LX16-CS324 FPGA. The machine is merely simulated with six states, which are selected using the traffic control method. Each state provides the appropriate delay, and the requisite traffic signals are turned ON/OFF for that delay. For example, just two routes are picked, and a control algorithm controls the traffic lights on those roads. This study introduced a flexible architecture that uses a clock divider to give a delay in a certain state and discusses the difficulty of modeling the FSM. The topic of selecting a state-encoding technique was also discussed.
A self-synchronization mechanism for aerial vehicles and air traffic control (ATC) systems was introduced in [70] for avoiding collision. It assists in obtaining a unique path for each aircraft in order to reduce conflict, but if this is not possible, the system gives an optimum option to utilize other shared resources.
An adaptive, flexible, and robust distributed TSC method based on spontaneous synchronization was introduced in [71], which occurs when local interaction leads to a stable state. In this method, every junction is represented by a phase-oscillator model. The global coordination is achieved by synchronizing these phase oscillators to maintain the platoons flow.
Adacher et al. [72,73] proposed a distributed algorithm for solving the synchronization problem of urban traffic signals. They calculated the weighted sum of delays at signalized intersections by CTM-UT; i.e., (Cell Transmission Model for Urban Traffic) via simulation. A surrogate method (SM) based on an online and stochastic control scheme has been used to synchronize the traffic signals.
An IoT-based traffic signal synchronization method was proposed in [74] to reduce the traffic congestion at road intersections. This method allows the vehicles to travel a long distance by minimizing the number of STOP and GO occurrences by adapting the A self-synchronization mechanism for aerial vehicles and air traffic control (ATC) systems was introduced in [70] for avoiding collision. It assists in obtaining a unique path for each aircraft in order to reduce conflict, but if this is not possible, the system gives an optimum option to utilize other shared resources.
An adaptive, flexible, and robust distributed TSC method based on spontaneous synchronization was introduced in [71], which occurs when local interaction leads to a stable state. In this method, every junction is represented by a phase-oscillator model. The global coordination is achieved by synchronizing these phase oscillators to maintain the platoons flow.
Adacher et al. [72,73] proposed a distributed algorithm for solving the synchronization problem of urban traffic signals. They calculated the weighted sum of delays at signalized intersections by CTM-UT; i.e., (Cell Transmission Model for Urban Traffic) via simulation. A surrogate method (SM) based on an online and stochastic control scheme has been used to synchronize the traffic signals.
An IoT-based traffic signal synchronization method was proposed in [74] to reduce the traffic congestion at road intersections. This method allows the vehicles to travel a long distance by minimizing the number of STOP and GO occurrences by adapting the traffic light phases at the road intersections ( Figure 5). This method reduces the average travel time up to 39% using traffic simulator SUMO in comparison to the other fixed time and non-synchronized traffic control strategies.
Nesmachnow et al. [75] introduced the synchronization of traffic lights using a parallel evolutionary algorithm for Bus Rapid Transit systems by considering various features such as real maps and mobility data and analyzing these features for optimization. This method improves the average speed of public transport up to 15.3% and other vehicles up to 24.8%. Using this method, different priorities can be assigned to buses and other vehicles using a multi-objective optimization analysis.
A model for synchronizing the traffic lights at road intersections is presented in [76]. This method uses a hybrid metaheuristic approach that combines variable neighborhood search and Tabu search into one algorithm. The method incorporates a memory structure into an iterative local search, allowing for a wider range of solutions. Certain changes such as modification of a mixed integer linear model's constraints and the generalization of some bounds for integer variables were made to the MAXBAND in this algorithm. traffic light phases at the road intersections ( Figure 5). This method reduces the average travel time up to 39% using traffic simulator SUMO in comparison to the other fixed time and non-synchronized traffic control strategies. Nesmachnow et al. [75] introduced the synchronization of traffic lights using a parallel evolutionary algorithm for Bus Rapid Transit systems by considering various features such as real maps and mobility data and analyzing these features for optimization. This method improves the average speed of public transport up to 15.3% and other vehicles up to 24.8%. Using this method, different priorities can be assigned to buses and other vehicles using a multi-objective optimization analysis.
A model for synchronizing the traffic lights at road intersections is presented in [76]. This method uses a hybrid metaheuristic approach that combines variable neighborhood search and Tabu search into one algorithm. The method incorporates a memory structure into an iterative local search, allowing for a wider range of solutions. Certain changes such as modification of a mixed integer linear model's constraints and the generalization of some bounds for integer variables were made to the MAXBAND in this algorithm.
In [77], the decentralized spatial decomposition of a network has been used for the synchronization of traffic signals at the road junctions. This method uses the platoon model to calculate the weighted sum of delays induced by signalized crossings. To get a spatial breakdown of the network, distributed consensus approaches were utilized. Using the distributed communication architecture, each semaphore distributes information with its neighbors in a totally dispersed manner, hence removing the need for a central authority. A surrogate technique is used for synchronizing the traffic signals given the subnetwork.
A self-synchronization paradigm for connected vehicles in traffic networks has been used in [78] to solve the coordination problem at junctions. For safe spacing between cars moving on the same route, coordinating safe crossing at junctions with competing flows, Figure 5. IoT-based mechanism for traffic signal synchronization using sensor deployment on adjacent junctions [74].
In [77], the decentralized spatial decomposition of a network has been used for the synchronization of traffic signals at the road junctions. This method uses the platoon model to calculate the weighted sum of delays induced by signalized crossings. To get a spatial breakdown of the network, distributed consensus approaches were utilized. Using the distributed communication architecture, each semaphore distributes information with its neighbors in a totally dispersed manner, hence removing the need for a central authority. A surrogate technique is used for synchronizing the traffic signals given the subnetwork.
A self-synchronization paradigm for connected vehicles in traffic networks has been used in [78] to solve the coordination problem at junctions. For safe spacing between cars moving on the same route, coordinating safe crossing at junctions with competing flows, a multilayer decentralized control method is utilized. The Kuramoto equation is used to synchronize the phase and frequency of the network's agents (vehicles). This approach considered the vehicles as oscillating agents instead of the traffic lights. This technique is classified as a decentralized approach since it is interconnected to achieve network-wide synchronization. It combines the advantages of coordinating vehicle crossings at particular crossroads with synchronizing flow from nearby junctions in this way.
In [79], floating car data (FCD) was used to synchronize the traffic signals at intersections. This approach used the penetration rate of "instrumented" vehicles to obtain synchronization in adaptive traffic light systems based on floating car data (FCDATS) in competition cooperation. This technology increases the traffic safety and energy efficiency of transportation systems.
Yang et al. [80] used cellular automaton for synchronizing traffic signals by building a network model of two intersections. By changing the signal duration and green time offset, the synchronization of the signals is investigated, and results are obtained in MATLAB. An optimal signal scheme reduces the overall delay and increases the traffic flow at road junctions. Next, the network's essential features, such as fundamental and time-space diagrams, are addressed.
Li et al. [81] introduced a time synchronization method among VANET devices, claiming that On-Board Units (OBUs) can synchronize with other OBUs or RSUs on their own initiative. This technique lowers the synchronization error to less than 0.3 ms when there is no center node in the Basic Service Set, which fulfills the accuracy criterion of the VANET specification. In the Basic Safety System (BSS), a time-synchronization mechanism is introduced in which there is no central node, resulting in a more stable network. This method can achieve nanosecond precision while avoiding estimate errors. This method can correctly implement time synchronization in a BSS. In addition, there is a topology for the whole VANET in order to achieve time synchronization among RSUs. As a result, the amount of handoff done by OBUs can be drastically decreased. It can also greatly increase the system's stability.
A survey of the synchronization techniques is presented in [82], which addressed various situations and current cellular network generations while reviewing synchronization standards and approaches for V2V and Device-to-Device (D2D) enabled cellular networks. They looked at the major channels, entities, and signals that V2V and D2D enabled cellular networks employ for synchronization in various circumstances. Furthermore, to highlight current developments and give future perspectives, the focus is on the synchronization in V2V and Vehicle to Everything (V2X).
Du et al. [83] presented a rule-based algorithm for V2X speed synchronization at road junctions. Vehicles entering from separate lanes form a virtual platoon. When traffic is light, this system allows for fewer unnecessary stops, resulting in energy savings and an increase in average speed. As a result, speed synchronization is no longer effective, as the volume of traffic increases and cars must come to a halt. Virtual platooning is expanded here by introducing some principles that enable a virtual platoon to dynamically modify its whole behavior in response to traffic flow growth while maintaining safety. This technique is extremely efficient in boosting average speed while maintaining the intersection's high capacity.
The significance of nearby junctions has not been studied in the above cited research works. The categorization of existing literature on synchronization of traffic signals is shown in Table A1 (Appendix A). These works are restricted to one intersection only, which makes the problem appearance more complex and extensive in scope. More efforts should be made in order to obtain maximal modeling, control, and monitoring of several synchronized junctions.

Discussion and Potential Directions
The primary goals of TSC technology research were to improve traffic safety, improve trip quality, and minimize pollution emissions. Traffic congestion has been a serious concern in metropolitan areas as vehicular traffic has increased. The problem is even worse at road intersections, where pre-timed traffic signals produce delays, increased vehicle activity costs, and air pollution. Delay is caused by the status of one light at a crossroad, which impacts the flow of traffic at neighboring intersections also. Additionally, the traditional traffic system does not account for mishaps, road closures, or vehicle breakdowns, which contribute to travel delays. Furthermore, a major issue is the seamless movement of emergency vehicles with greater demands, such as emergency vehicles, ambulances, fire brigade, police, and VIPs, through crossing points. As a result, the regular traffic system must be upgraded in order to alleviate traffic congestion and minimize waiting times and travel time as well as maximize vehicle safety, efficiency, and economic benefits [84].
The majority of traffic monitoring technologies at junctions rely on high-cost cameras and sensors that are difficult to install and maintain. Computationally intensive traffic signal control technologies are utilized at road junctions. As a result, real-time signal control tactics at road crossings must be enhanced in order to eliminate traffic congestion. A massive disadvantage is that the traditional methods do not synchronize the traffic signals of a region.
To address these concerns, traffic light synchronization is a technique in which a vehicle traveling along one side of the road at a specific speed can continue to the opposite end of the road indefinitely by obtaining the maximum number of green lights at all the intersections. Traffic signal synchronization is required so that the green light cycles for a series of junctions can be synchronized, enabling the greatest number of automobiles to pass through while avoiding waits and delays. Synchronization can be done at different levels. Figure 6 shows traffic signal synchronization at level I and level II. We can define the levels of synchronization within an area. One will be the inner circle and another will be the outer circle, where the inner circle reflects peer junctions of level I and the outer circle reflects the peer junctions of level II. Level I consists of the nearest neighboring junctions, and level II consists of the junctions falling in the external area. The intelligence at level II also takes into consideration the density of the adjacent junctions. These levels of synchronization will help in reducing the congestion at neighboring intersections. In this way, the system will become smart when it is connected to the server. bulances, fire brigade, police, and VIPs, through crossing points. As a result, the regular traffic system must be upgraded in order to alleviate traffic congestion and minimize waiting times and travel time as well as maximize vehicle safety, efficiency, and economic benefits [84].
The majority of traffic monitoring technologies at junctions rely on high-cost cameras and sensors that are difficult to install and maintain. Computationally intensive traffic signal control technologies are utilized at road junctions. As a result, real-time signal control tactics at road crossings must be enhanced in order to eliminate traffic congestion. A massive disadvantage is that the traditional methods do not synchronize the traffic signals of a region.
To address these concerns, traffic light synchronization is a technique in which a vehicle traveling along one side of the road at a specific speed can continue to the opposite end of the road indefinitely by obtaining the maximum number of green lights at all the intersections. Traffic signal synchronization is required so that the green light cycles for a series of junctions can be synchronized, enabling the greatest number of automobiles to pass through while avoiding waits and delays. Synchronization can be done at different levels. Figure 6 shows traffic signal synchronization at level I and level II. We can define the levels of synchronization within an area. One will be the inner circle and another will be the outer circle, where the inner circle reflects peer junctions of level I and the outer circle reflects the peer junctions of level II. Level I consists of the nearest neighboring junctions, and level II consists of the junctions falling in the external area. The intelligence at level II also takes into consideration the density of the adjacent junctions. These levels of synchronization will help in reducing the congestion at neighboring intersections. In this way, the system will become smart when it is connected to the server.  The designed traffic signal synchronization framework can work efficiently with any technology, whether it is DSRC, image processing, sensors, or any other technology that is used to collect traffic density at intersection. With minor changes to the current system, new junctions can also be incorporated for synchronization. In this way, it is possible to extend this system for covering all the intersections of a city or state, thus moving toward a well-coordinated and lucid method of traffic control.

Conclusions
Traffic signal synchronization for intelligent vehicles is a revolutionary innovative technology with exceptional potential in the intelligent transportation systems. Hence, this article provides a comprehensive survey on traffic light synchronization techniques for intelligent vehicles. First of all, we outline the different traffic light control technologies used for congestion control at intersections. Then, typical applications of synchronization of traffic signals are discussed. Next, we summarize the development status of synchronization of traffic signals for intelligent vehicles by surveying all the research works on traffic signal synchronization. In addition, we discuss many emerging and potential directions of traffic light networking. This study will provide researchers with the basis for more understanding and investigation of the synchronization of traffic lights for intelligent vehicles.
Author Contributions: I.T.: Lead author of manuscript (Corresponding Author), Conceptualization and methodology, writing-original draft preparation, investigation, editing; I.S. and N.P.: Second author, research design, guidance, and reviewing. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
Acknowledgments: This work has been carried out under the supervision of Indu Sreedevi and Neeta Pandey, Department of ECE, Delhi Technological University, and I (Ishu Tomar) pay immense gratitude toward my guides for enlightening me through the process. In addition, I (Ishu Tomar) would like to acknowledge the financial assistantship provided by Delhi Technological University.

Conflicts of Interest:
The authors declare that they have no conflict of interest.