پژوهشگاه علوم و فناوری اطلاعات ایران (ایرانداک) پژوهشنامه پردازش و مدیریت اطلاعات 2251-8223 40 ویژه نامه انگلیسی 4 2025 07 23 Adaptive AI-Driven Network Slicing in 6G for Smart Cities: Enhancing Resource Management and Efficiency Adaptive AI-Driven Network Slicing in 6G for Smart Cities: Enhancing Resource Management and Efficiency 1541 1573 728445 10.22034/jipm.2025.728445 FA Nameer Hashim Qasim Al-Rafidain University College Baghdad, Iraq,10064 0000-0002-7283-0594 Omar Chafic Emirates Aviation University, Dubai 686, United Arab Emirates 0009-0009-3419-1662 Journal Article 2025 08 17 <strong>ABSTRACT </strong><br /><strong>Background:</strong> Smart city evolution is fast-paced, and imposes severe demands on telecom infrastructures: it must be highly flexible and scalable for coping with bursty traffic loads and heterogeneous service needs. Legacy network systems are not well suited to handle the changing requirements of smart city environments with autonomous cars, IoT, and public safety systems.<br /><strong>Objective</strong>: The study to offer an AI-native network slicing framework for 6G smart city networks in order to improve dynamic resource control and management. The framework aims to enhance the delay, energy, and resource performance metrics which are significant for smart city services.<br /><strong>Method:</strong> To facilitate the real-time network resource orchestration depending on the changing traffic requirements and user preferences, the authors consider moving target defense adapted artificial intelligence with a Deep Reinforcement Learning (DRL) model. Simulations were carried out to compare the AI-native model to conventional and AI-supported slicing methods.<br /><strong>Results</strong>: Simulation results validate that the AI-native network slicing framework outperforms current 5G solutions with 25% reduction in latency and 20% increase in energy efficiency. Furthermore, the model's online resource allocation scheme can enhance the utilization efficiency of the bandwidth and the energy by 15% compared with the traditional approaches. Such improvements especially in critical applications like traffic management, emergency response, and health care would be important.<br /><strong>Conclusion:</strong> The presented results demonstrate that AI-native network slicing is a viable, flexible, and scalable solution for 6G smart city networks. The framework is designed to support the future sustainable and high-performance requirements of urban infrastructures, providing both energy-efficient real-time adaptability. This study provides an overarching front-to-end outlook to address the management issues of sophisticated resource systems, and puts AI-native network slicing at the base level of the emerging smart cities. <strong>ABSTRACT </strong><br /><strong>Background:</strong> Smart city evolution is fast-paced, and imposes severe demands on telecom infrastructures: it must be highly flexible and scalable for coping with bursty traffic loads and heterogeneous service needs. Legacy network systems are not well suited to handle the changing requirements of smart city environments with autonomous cars, IoT, and public safety systems.<br /><strong>Objective</strong>: The study to offer an AI-native network slicing framework for 6G smart city networks in order to improve dynamic resource control and management. The framework aims to enhance the delay, energy, and resource performance metrics which are significant for smart city services.<br /><strong>Method:</strong> To facilitate the real-time network resource orchestration depending on the changing traffic requirements and user preferences, the authors consider moving target defense adapted artificial intelligence with a Deep Reinforcement Learning (DRL) model. Simulations were carried out to compare the AI-native model to conventional and AI-supported slicing methods.<br /><strong>Results</strong>: Simulation results validate that the AI-native network slicing framework outperforms current 5G solutions with 25% reduction in latency and 20% increase in energy efficiency. Furthermore, the model's online resource allocation scheme can enhance the utilization efficiency of the bandwidth and the energy by 15% compared with the traditional approaches. Such improvements especially in critical applications like traffic management, emergency response, and health care would be important.<br /><strong>Conclusion:</strong> The presented results demonstrate that AI-native network slicing is a viable, flexible, and scalable solution for 6G smart city networks. The framework is designed to support the future sustainable and high-performance requirements of urban infrastructures, providing both energy-efficient real-time adaptability. This study provides an overarching front-to-end outlook to address the management issues of sophisticated resource systems, and puts AI-native network slicing at the base level of the emerging smart cities. https://jipm.irandoc.ac.ir/article_728445_314092559e96a0d7ffad961ba9dc94a7.pdf