چکیده مقاله
The IP Multimedia Subsystem IMS serves as the core architecture for real time multimedia services in 5G and beyond Edge IMS is a key evolution in modern telecom networks, where architectural components are moved closer to the user for improved efficiency With traditional architectures, IMS scalability and energy efficiency are challenged by signaling storms, HSS latency, and resource over provisioning in edge deployments This paper proposes a novel AI augmented IMS architecture based on Docker Swarm—which can act as a lightweight, low overhead orchestration platform ideal for edge environments We integrate four AI driven modules: 1 LSTM based horizon aware predictive scaling of S CSCF instances, preemptively adjusting replicas based on SIP session trends; 2 XGBoost powered dynamic routing of Cx/Diameter queries to geo sharded HSS replicas to minimize lookup latency; 3 1D CNN autoencoder for temporal pattern preserving anomaly detection in real time RTP streams; and 4 multi objective deep reinforcement learning DDPG balancing energy consumption, SLO compliance, and placement constraints during CNF orchestration These modules operate within a Docker Swarm cluster, leveraging native VXLAN overlays and service APIs for closed loop automation Evaluated in a simulated smart city edge deployment with 500K subscribers, our framework reduces SIP transaction latency by 45%, increases call density per watt by 35%, and achieves 95% accuracy in preemptive fault prediction The architecture aligns with 3GPP Release 17 and ETSI NFV standards, offering a low footprint, API consistent, and operationally simpler alternative to Kubernetes for cloud native IMS at the edge
کلیدواژهها
نویسندگان
Ali Barkati
Department of Electrical Engineering Islamic Azad University of Mahshahr Telecommunication Company of Iran
Mohammad Javad Rashti
Department of Computer Engineering Shahid Chamran University of Ahvaz
Pourya Barkati
Department Of Computer Engineering Bu Ali Sina University
شیوه ارجاع
Barkati, Ali and Rashti, Mohammad Javad and Barkati, Pourya,1404,AI-driven IMS-at-the-Edge using Docker Swarm: Optimizing Signaling, Traffic Management, and Fault Prediction,The 4th International Conference and the 9th National Conference on Computers, Information Technology and Artificial Intelligence Applications,Ahvaz
ارائهشده در
مجموعه مقالات پنجمین همایش ملی و نخستین همایش بین المللی محاسبات نرم علوم مهندسی در صنعت و جامعه26 بهمن 1404 · ایرانشهر