In the era of E-Learning, most research on learner model is focused on the analysis of learning-related data, but ignored the analysis of the multidimensional data and relationship among them. This will make it difficult to constructing dynamic, real-time and accurate learner model. However, in the era of U-Learning, the learning device has powerful ability of sensing the context awareness. With these context-awareness technologies, the U-Learning system can collect multi-sensor data intelligently, and then analyze these data by ubiquitous computing, providing learners with timely and accurately personalized study guide. With the wide use of context-awareness technology in the education field, this paper discusses how to use context-aware technology to design dynamic and accurate individualized learner model. Firstly, we describe the contextual factors and classification involved in the U-Learning environment, and then analyze the relationship of contextual factors and learner characteristics. Secondly, we choose an appropriate algorithm to analyze the multi-sensor data which sense and identify learner context automatically, discussing the design of dynamic and real-time learner model by pattern matching and context reasoning. Finally, we propose a framework of learner model which is based on the analysis result of context-aware computing, and then focus on the design prototype system to enhance U-Learning system of adaptive ability.