Mobile Interaction Technology-Driven Blended Learning for Higher Education

Abstract

In blended learning environments in higher education, the use of mobile devices has long been constrained to content delivery and superficial interaction, whereas the multimodal sensing and edge-computing capabilities of smartphones have not been fully exploited. A mobile interaction technology-driven blended learning model was proposed, in which a device–edge–cloud collaborative federated learning architecture was constructed to enable local sensor feature extraction and differential privacy perturbation on smartphone terminals, with only desensitized gradients being transmitted to edge gateways. To achieve unobtrusive learning-state perception, a posture-recognition pipeline integrating complementary filtering and a lightweight MobileNetV3 model was designed. Six classroom postures were classified in real time using the accelerometer, gyroscope, and magnetometer. In addition, intergroup distance was dynamically estimated through Bluetooth Received Signal Strength Indicator (RSSI) measurements combined with Kalman filtering. When the physical distance between learners remains below 2 m for more than 30 s, vibration notifications and personalized micro-quizzes are automatically triggered by the edge intelligence engine, with quiz difficulty adaptively adjusted according to the exponentially decayed historical accuracy rate of each learner. Furthermore, long short-term memory networks were employed to model touch pressure, screen states, and posture sequences, thereby enabling the adaptive delivery of post-class augmented-reality exercises. Multipath transmission control protocol (MPTCP) and Q-learning-based path scheduling were incorporated to support seamless session migration between classroom and outdoor learning scenarios. Experimental validation demonstrates that a high-accuracy and low-latency mobile interaction framework can be achieved using only native smartphone sensors. A practical and scalable technical solution is therefore provided for blended learning for ubiquitous higher education environments.