Towards Smarter Cities: An Adaptive AIoT Solution with Dynamic Sensor Group Selection for Real-Time Traffic Monitoring and Energy Efficiency

Efficient traffic monitoring and safety in smart cities demand real-time adaptability and energy-efficient operations. Static sensor configurations fall short in dynamic urban environments, leading to excessive energy use and delayed response times. This paper presents an adaptive Artificial Intelligence of Things (AIoT) framework with dynamic sensor group selection for intelligent traffic safety. Our system employs a context-aware sensor selection algorithm that activates only the most relevant sensors based on real-time conditions, including weather, traffic density, and time of day, thereby reducing unnecessary sensor usage and conserving energy. The framework's incorporation of weighted sensor prioritization and dynamic activation probabilities enhances its adaptability, achieving significant energy savings and accuracy improvements in real-time traffic monitoring scenarios. Energy management layer incorporates that dynamically adjusts power allocation and employs duty cycling to prioritize critical sensors while minimizing energy consumption. To address latency requirements, we leverage edge computing and vehicle-to-vehicle (V2V) communication, processing critical data locally and transmitting urgent alerts directly between vehicles. Additionally, low-latency communication protocols enable efficient data transfer for non-critical information to central servers. Experimental evaluations show that the proposed adaptive AIoT system reduces energy consumption by up to 30% and improves response times by 70% compared to non-adaptive configurations. These results demonstrate the potential of adaptive AIoT with dynamic sensor group selection to enhance urban traffic safety, providing a scalable, responsive solution for real-time smart city applications.