Research is now shifting away from Virtual Learning Environments (VLEs) and towards the use of the Personal Learning Environment (PLE). A review of a number of PLE architectures are presented in the literature, and while they convey well the concept of a PLE, nevertheless they could best be described as high-level architectures, (sometimes referred to as frameworks in the literature), which focus mainly the functionality of PLEs. In particular, there is little published which gives a detailed designed of a PLE architecture. Moreover, the published work focuses largely on the support for lifelong learning and formal / informal learning; these are two of the main limitations of VLEs. However, this study argues that unexplored potential remains, as there is scope for PLEs to cover more areas. To the best of our knowledge, none of the existing PLE architectures have context-aware systems embedded within their architecture. There is no intelligence in these architectures to filter the e-resources and to predict the user need. In addition, the current PLE architectures are not dynamic; it cannot adopt the user current situation. The user of the current PLE architectures receives too much e-resource. The architecture proposed in this research incorporates a context-aware engine. Thus there is intelligence built into the architecture and thus the PLE system is automatically responsive to the context information. There are three types of sensors in any context-aware system (physical, virtual and logical), and these are the elements of the system that gather the context information. In this research, the emphasis will be on virtual sensors which gather the information from virtual space; virtual space here includes any systems which produce information as a set of results. Thus, the context-aware architecture and the implementation of the context-aware engine are major contributions of the work.