面向个性化学习的应用需求，本项目拟构建基于教育云脑、边缘网络和人机物终端所构成的信息-物理-社会三元融合系统，聚焦研究边云协同关键技术，全面深度感知学生的学习状态和效果，实现低时延的高密度无线联网传输，完成学习过程实时建模分析和异构计算加速，提供多种人工智能支持的智慧导学服务。本项目拟设计和实现支持学习全过程协同感知的软件定义测量系统；面向基于连接高密度部署的异构随机接入网络，提出一种基于双队列的嵌入式离散时间马尔可夫过程模型，分析和优化随机接入网络下随机接入时延性能，并通过实际网络测量对该理论模型进行性能评估；设计和实现融合多模态数据（包括学习者的生理数据和学习过程视频数据）的学习推荐系统，并基于边云的协同机制和可编程边缘节点的异构计算能力对复杂的学习过程视频分析算法进行加速。本课题的研究成果，将对于构建人工智能驱动的智慧学习环境，显著提升学生学习体验，具有重要的现实意义。

For the future personalized learning applications, in this project we propose to construct a cyber-physical-social system based on the education cloud, edge networks and end-devices. We aim to study the cooperation schemes of the proposed edge-cloud hybrid architecture, including deep sensing the whole-learning process of students, achieving low delay in dense random access networks, cognitive modeling and heterogeneous computing, and providing artificial-intelligence-supported learning guidance. We plan to design and implement a software defined measurement system to sense the students’ learning process from multi-modal data sources collaboratively. For the densely-deployed random access networks, based on a novel two-queue embedded discrete Markov Chain, we propose a unified model to analyze and optimize the delay performance. This model will be validated and evaluated using testbed measurements. Last, we plan to design and implement a learning recommendation system based on multi-modal data sources including wearable data and learning videos. The complex video analytics can utilize the cooperation schemes of the proposed edge-cloud architecture and the heterogeneous computing modules for significant acceleration. Our research may provide insights into constructing a future intelligent learning environment for enriched personalized experiences.