Bridging Standardization and Intelligence in oneM۲M-Based IoT: A Systematic Review, Taxonomy, and Research Roadmap

The growing scale and heterogeneity of Internet of Things (IoT) ecosystems have made intelligent, standardized service management essential, positioning oneM۲M as a horizontal service layer capable of mediating interoperability while artificial intelligence (AI) increasingly governs quality-of-service control, security, resource allocation, and predictive analytics across connected devices. Despite substantial research activity at this intersection, existing studies remain fragmented across disconnected AI paradigms and application domains, lacking a unified taxonomy capable of positioning classical machine learning, deep learning, federated learning, and edge-intelligence contributions relative to the standardized oneM۲M architecture they target. This survey addresses that gap through a systematic review of forty Q۱-indexed studies published between ۲۰۱۸ and ۲۰۲۶, analyzed through a novel four-tier taxonomy spanning standardization and interoperability foundations, distributed intelligence paradigms, functional optimization objectives, and cyber-physical domain realization. This study analyzes the reviewed corpus through statistical characterization of research clusters and technique distributions, comparative evaluation of AI methodologies against reported datasets, evaluation metrics, and architectural integration levels, and structured identification of recurring challenges and research gaps. This work identifies that classical ensemble methods embedded within CSE-native MAPE-K control loops represent the most architecturally mature research direction, while federated learning and digital-twin research, despite considerable algorithmic sophistication, exhibit minimal integration with the oneM۲M standard itself; concurrently, standardized benchmarking, explainability, and security-aware intelligence remain consistently underdeveloped across nearly all research categories. These findings motivate future research directions oriented toward unifying resource-efficient, privacy-preserving, explainable, and security-aware intelligence within native oneM۲M services rather than as isolated methodological demonstrations. By consolidating a fragmented literature into a coherent analytical framework, this survey provides a foundation for advancing standards-integrated, trustworthy, and scalable AI-enabled IoT research.

کلیدواژه ها:

oneM۲M ، Internet of Things (IoT) ، Artificial Intelligence ، Machine Learning ، Deep Learning ، Federated Learning ، Edge Intelligence ، Survey