Brief Review of Using Machine Learning for Traffic Engineering in Software-Defined Networks
Entisar H. Khalifa · Pakistan Journal of Life and Social Sciences (PJLSS) · 2025
This review is a brief exploration of deploying Machine Learning (ML) in Traffic Engineering (TE) for Software Defined Networks (SDN).SDN changes traditional network management by separating the control plane from the data plane, opening up new possibilities for flexible and adaptive traffic control.As we show, TE in SDNs can optimize network performance by using resources more efficiently, cutting down on latency, and reducing congestion-all while responding to real-time conditions to maintain high Quality of Service (QoS).However, taking full advantage of these benefits requires advanced algorithms and real-time data analysis, which can be computationally demanding.TE also relies on having accurate, up-to-date information about the network.Meanwhile, ML is making SDNs more effective by integrating with technologies like Edge Computing, Network Function Virtualization (NFV), and the Internet of Things (IoT).This combination enables real-time analytics, quick decision-making, intelligent routing, load balancing, and stronger security.Still, these integrations bring fresh challenges in scalability and interoperability, meaning we need major investments in both infrastructure and expertise.Even with all the progress made so far, several hurdles remain.These include issues with scaling up, maintaining robust security, and making split-second decisions in real-time.Looking ahead, future research should concentrate on autonomous networking, energy-efficient ML techniques, and hybrid ML solutions, aiming to reach new heights in network security and performance.