Primary education environments use mobile networks for student devices, tablets, and educational IoT systems

Mobile and IoT-enabled technologies are increasingly vital in education, where personalized learning, real-time flexibility, and QoS are increasingly vital. However, the diverse devices and latency-sensitive applications, as well as the nature of educational applications, render static network management unsuitable for modern digital classrooms. This study proposes LQ-EduNet, an AI-powered framework designed to enhance personalized learning to improve mobile educational network QoS and user experience. The proposed system integrates an LSTM network to model user behavior and content access sequences across time. A Deep Q-Network (DQN) controller adaptively orchestrates network resources in real-time. The LSTM module creates high-dimensional embeddings of student interactions to instruct the edge DQN agent which optimises bandwidth, latency, and spectrum allocation as the network advances. Efficacy, individualization, and equity are considered via a composite reward function that guides learning. The framework was tested on 100 student devices and 40 IoT sensors in a simulated 5G classroom. LQ-EduNet outperformed baseline models in end-to-end latency (39.2%), tailored content delivery accuracy (31.5%), and throughput (44.8%). while LQ-EduNet shows an apparent improvement from ~ 65% to ~ 95% across 10–100 students compared to SEM-ANN (~ 50–75%), WAD-YOLOv8 (~ 40–60%), and IDBN-CNN (~ 30–45%),Edge-level execution safeguards data privacy and system resilience in mobile devices, particularly in conditions of bursty traffic. In conclusion, LQ-EduNet provides a smart and scalable solution to enhancing and customizing real-time quality of service in educational mobile communication settings.