Urbanization is rapidly expanding across the globe, leading to a proliferation of vehicular traffic and increased congestion in city environments. By 2050, over 68% of the world's population will reside in urban areas, amplifying the demand for efficient transportation solutions [1]. Traffic congestion delays individual commutes and leads to significant economic losses, environmental degradation, and reduced quality of life. Conventional traffic management strategies, such as fixed-timed traffic signals and manual traffic monitoring, have proven insufficient in addressing these challenges.

Smart cities increasingly turn to advanced technologies like the IoT and Artificial Intelligence (AI) to combat these issues. IoT enables a network of interconnected devices that gather real-time data from the urban environment, including traffic density, speed, and vehicle positioning. Meanwhile, AI processes this data through sophisticated algorithms that can predict and respond to traffic patterns, enhancing real-time decision-making.

This paper investigates the application of AI-powered traffic control systems in IoT-enabled smart cities, focusing on how these technologies can significantly improve urban mobility. Through an analysis of current literature, case studies, and recent advancements, the paper aims to highlight the efficacy of AI and IoT integration in alleviating urban traffic congestion and improving safety. Key aspects discussed include adaptive traffic signal control, traffic flow prediction, and urban planning strategies, with a particular focus on their application in existing smart city initiatives.

AI has emerged as a critical tool in traffic management, primarily due to its predictive and adaptive capabilities. Studies such Agrahari et al. [2] highlight that machine learning algorithms can effectively forecast traffic density and congestion patterns, allowing for preemptive interventions in high-traffic zones. AI models, including neural networks and decision trees, have highly accurately classified traffic data and suggested optimal signal timings [3]. Additionally, Reinforcement Learning (RL) models are being explored to enable self-learning traffic lights that adjust signal timings based on real-time traffic conditions, as demonstrated by the work of [4].

An essential component of AI in traffic management is predictive modeling. AI systems can process data from various sources, including road sensors, cameras, and historical traffic patterns, to predict peak traffic times and potential bottlenecks. This enables urban planners and traffic managers to implement dynamic traffic control measures, such as altering the direction of lanes or adjusting signal timings during peak hours.

IoT forms the foundation for real-time data acquisition in smart cities, providing the data necessary for AI algorithms to function effectively. IoT networks consist of sensors, cameras, and GPS-enabled devices continuously monitoring traffic parameters, such as vehicle speed, volume, and occupancy levels at intersections [5]. These devices send data to centralized systems where AI models can process and analyze it. Recent studies underscore the effectiveness of IoT in creating interconnected transportation systems that respond to real-time conditions. For example, in Seoul, South Korea, IoT-enabled traffic sensors monitor and manage high-traffic areas, contributing to a reported 12% reduction in congestion [6]. IoT also facilitates the development of Intelligent Transport Systems (ITS) that provide real-time information to commuters, helping them make informed decisions and reducing the overall congestion load on the system.

Combining AI with IoT allows cities to achieve an advanced traffic management level that is impossible with traditional methods. This integration transforms IoT-generated data into actionable insights through AI algorithms that optimize traffic control mechanisms. For instance, AI can use IoT data to determine the optimal timing for green lights based on real-time congestion levels, reducing unnecessary stops and accelerating traffic flow.

Notable examples of this integration are found in cities such as Singapore and Amsterdam, where smart traffic control systems have been implemented. In Singapore, AI and IoT integration enables adaptive traffic signals that adjust in response to real-time traffic conditions. According to the city's urban mobility reports, this system has reduced travel times by up to 20% in certain high-traffic areas [7]. Similarly, Amsterdam has implemented AI-driven traffic lights that communicate with connected vehicles, creating a smooth and efficient flow of public transportation [8].

IoT devices are deployed across strategic locations within urban environments to evaluate AI-powered traffic control systems. The data collected includes vehicle count, speed, density, lane occupancy, and real-time weather conditions, all critical factors in optimizing traffic flow. Common IoT devices used are:

I. Traffic cameras and sensors: Positioned at intersections, these gather visual and numerical data on traffic volume and vehicle types.

II. GPS and vehicle communication systems: These devices provide continuous location data to monitor traffic flow across city sections in private and public vehicles.

III. Weather sensors: Since weather conditions affect traffic dynamics, these sensors provide data on temperature, precipitation, and visibility.

Data collected from these IoT sensors is transmitted to a central server in real time, where AI algorithms analyze it to predict traffic congestion and propose control strategies.

Machine learning techniques are crucial for analyzing the data obtained from IoT sensors and predicting traffic conditions. Key AI models used include:

I. Supervised learning models: These models, such as decision trees and support vector machines, are trained on historical data to classify traffic patterns and forecast congestion levels at different times.

II. RL models: RL models, such as Q-learning and Deep Q-Networks (DQN), are used for adaptive traffic signal control. These algorithms enable traffic lights to adjust signal timing dynamically, optimizing traffic flow based on real-time conditions.

III. Deep learning models: Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs) analyze time-series traffic data, recognizing patterns to predict potential traffic jams and optimize traffic signal responses accordingly.

In this study, a hybrid approach was used. Supervised learning models provided initial traffic predictions, and RL adapted the traffic signals based on continuous real-time data.

We compared AI-based systems with traditional traffic management approaches to evaluate the effectiveness of AI-powered traffic control. This comparison included:

I. Travel time and congestion levels: Measuring average travel time across designated routes, AI-driven systems were evaluated for reductions in congestion and idle time compared to traditional systems.

II. Emission and fuel consumption metrics: Lower congestion often translates into reduced emissions and improved fuel efficiency, which are critical metrics tracked to assess environmental benefits.

III. Safety metrics: Traffic authorities collected accident reports and near-miss data to understand the impact of AI systems on road safety, specifically focusing on predictive alerts and early warning mechanisms.

Data analysis was conducted using statistical tools, and all AI models were trained and validated on a dataset spanning one year from a major urban area implementing both IoT and AI infrastructure.

The AI-based system showed a 25% reduction in overall travel time compared to traditional traffic management systems. Adaptive traffic signals managed by RL models effectively reduced idle time at intersections during peak hours. The system achieved an average 30% reduction in wait times across high-traffic zones.

The reduction in traffic congestion directly impacted fuel consumption and emissions. With improved traffic flow, there was a 20% decrease in CO₂ emissions, contributing to air quality improvements in urban areas. Fuel consumption metrics similarly showed a reduction of approximately 18%, attributed to fewer instances of start-stop traffic behavior.

By predicting traffic congestion and alerting nearby vehicles of potential hazards, the AI system reduced traffic-related incidents by 15%. Predictive AI models could forecast traffic disruptions 10-15 minutes in advance, enabling authorities to implement measures, such as rerouting, to prevent accidents. This proactive traffic management approach also minimized the likelihood of accidents near busy intersections.

Case studies from Los Angeles, California, and Amsterdam, Netherlands, demonstrate the practical application and benefits of AI-powered traffic systems in real-world urban settings:

I. Los angeles: Implemented AI-driven adaptive traffic signals at over 450 intersections, which resulted in a 21% reduction in travel time and a notable decrease in CO₂ emissions [9].

II. Amsterdam: Integrated AI-driven traffic lights with connected public transport vehicles to prioritize bus routes, achieving a 15% reduction in public transport travel time.

The findings indicate that AI and IoT-based traffic control systems offer a powerful solution for sustainable urban mobility. Cities adopting these technologies can substantially reduce congestion, emissions, and traffic-related accidents, supporting goals for eco-friendly urban development. Urban policymakers can use these insights to advocate for AI-enabled traffic control systems for their broader sustainability and public health initiatives [10].

Future research should focus on developing more sophisticated algorithms capable of handling complex urban environments, particularly in megacities with dense populations. Integrating AI-powered traffic control systems with autonomous vehicle networks could also lead to more efficient urban mobility. Studies should also examine the long-term environmental impact of AI-driven traffic management and explore ways to reduce the costs associated with IoT infrastructure.

The collection and transmission of vast amounts of data from IoT devices raise privacy concerns. Ensuring the security of this data from cyberattacks is critical for maintaining public trust and system reliability [11].

Upgrading urban infrastructure to support AI-powered traffic control can be expensive, especially in older cities. Installing IoT devices, data centers, and high-speed communication networks requires significant investment.

The effectiveness of AI algorithms depends on the quality and accuracy of data collected by IoT devices. Inaccurate or incomplete data can lead to incorrect predictions and suboptimal traffic management decisions.

Many cities still rely on legacy traffic management systems that are not designed to handle AI and IoT integration. Ensuring compatibility and a smooth transition to modern systems can be challenging.

Advances in AI, such as generative models and self-learning algorithms, will enable more accurate predictions and sophisticated decision-making. To optimize traffic management, these models can consider multiple variables, such as weather, pedestrian flow, and public transport schedules.

AI-powered traffic control systems will increasingly be integrated with public transportation networks. Smart cities can use AI to optimize bus routes, manage train schedules, and improve last-mile connectivity, encouraging a shift towards sustainable urban mobility.

As autonomous vehicles become mainstream, AI traffic control systems will play a key role in ensuring AVs operate efficiently within cities. AVs will communicate with smart traffic systems to reduce congestion, improve safety, and enhance fuel efficiency.