Alex Quantum Dilated Convolutional Neural Network–Based Framework for Reliable End‐to‐End Real‐Time Data Transmission and Routing in Software‐Defined Networking

ABSTRACT The traffic management in the network, which is able to serve latency‐sensitive applications such as video conferencing in a very efficient manner, is provided by a software‐defined networking (SDN) system architecture that is centralized and programmable. On the other hand, regular SDN methods typically show that the adaptations to dynamically changing network situations are too conservative and slow. In addition, as a result of insufficient management of long‐range traffic dependencies, they frequently suffer from jitter, latency spikes, and packet loss. To address these limitations, this paper proposes an intelligent SDN‐based routing framework employing a hybrid Alex Quantum Dilated Convolutional Neural Network (AQDCNN). The proposed AQDCNN combines the spatial–temporal feature extraction capability of AlexNet with the long‐term dependency modeling and quantum‐enhanced computational efficiency of QDCNN. The framework is implemented within a multicontroller SDN environment using Mininet and OpenFlow, enabling dynamic network endpoint provisioning, QoS‐aware traffic prioritization, and fault‐tolerant failure adaptation. The experimental results show that the model outperforms the tested algorithms in terms of minimum setup and fast recovery time (1.90 and 2.10 ms) with low recovery overhead (2.70 ms), as well as high FTI equal to 8.50 and the lowest path change frequency equal to 4.20, which are considered significant stability and service continuity results for experiments conducted on a real‐life network topology. The proposed AQDCNN framework is against existing CNN, GN‐DQN, DRL‐SDN, IFRA‐GLB, and MTF‐WMSSA–based routing methods, demonstrating superior results in fault tolerance, rerouting time, and flow setup efficiency.