Published Online:February 2026
Product Name:The IUP Journal of Telecommunications
Product Type:Article
Product Code:IJTC010226
DOI:10.71329/IUPJTC/2026.18.2.7-36
Author Name:Sakthi Saravanakumar P and V Shanmuganeethi
Availability:YES
Subject/Domain:Engineering
Download Format:PDF
Pages:7-36
The paper presents a systematic review of AI-driven traffic on the network classification schemes, AI-enhanced traffic engineering and optimization in software-defined networking (SDN), and self-healing, fault-tolerant SDN architectures. Fifty-one primary studies gathered from six literature databases and enriched by snowballing are classified and compared using a novel five-perspective taxonomy encapsulating SDN planes, learning paradigms, network functions, deployment loci and autonomy levels. The surveyed approaches are critically contrasted with respect to scalability, computational complexity, controller overhead, latency, adaptability, interpretability, energy footprint, fault tolerance, and deployment feasibility. In addition, the review positions itself against ten existing representative studies, identifies the research gap it fills, and examines emerging directions, including graph neural networks (GNNs) for routing, large language models (LLMs) for network management, digital twin networks (DTNs), intent-based networking, zero-touch network and service management, safe and multi-agent reinforcement learning, network foundation models, knowledge-graph-based management, and AIOps. The study concludes with a phased research roadmap toward intelligent, autonomous, and resilient SDN-based networks.
Modern communication networks were not designed around predictable traffic patterns. Contemporary dynamic platforms, including cloud computing, virtualization, and data-driven services, must cope effectively with fluctuating workloads, diverse application demands, and consistently stringent service-level requirements.