Abstract

The advancement of mobile sensing devices and mobile computing technologies have triggered new research opportunities in mobile edge sensing and computing, including activity recognition, wellbeing monitoring, user authentication, human dynamics tracking, etc. However, research in mobile edge sensing and computing suffers from labor-intensive training, unrealistic experimental environments, heavy environmental interferences in practical scenarios. In addition, different research groups usually conduct small-scale experiments separately, which makes it difficult to share the research results and data among groups in the same community. The U.S. mobile edge sensing and computing community demands an experimental infrastructure to share data/models nationwide and perform practical and repeatable experiments. The goal of this project is to build a large-scale, mobile edge sensing and computing infrastructure to provide practical experimental environments, rich user tools and services, and data/model sharing. Based on the proposed infrastructure, individual research groups can be connected to conduct large-scale research with low efforts. Many interdisciplinary communities can also be brought together, conducting research via the proposed infrastructure, including deep learning-based hardware design, smart healthcare, AR/VR, human flow monitoring, smart home, and smart city. The research results can benefit interdisciplinary curriculums with new research topics and tasks for undergraduate/graduate and minority students.

The proposed research infrastructure includes three organically connected functionalities to provide repeatable experimental environments, facilitate data/model-sharing, and join separated research groups on a national scale. In particular, this project develops mobile sensing functionalities for supporting compelling research in low-effort large-scale sensing data collection, robot-enabled experimenting, and privacy-preserved learning on mobile edge devices. Furthermore, this project develops an edge computing functionality integrating remote-operated mobile edge devices and mobile development kits to support research in software and hardware co-design and on-device AI learning for low-cost mobile devices. In addition, a novel data and model sharing functionality is developed to support a broad spectrum of mobile edge sensing and computing research areas. A uniform web portal is provided to allow users to use these functionalities remotely. The proposed infrastructure provides an essential hardware and software foundation that enables cutting-edge research in CISE research focuses, including mobile edge sensing, hardware and software co-design, and distributed computing with sharable large-scale data from practical environments. The outcome from this project, including the unique integrated functionalities, powerful tools and services, and comprehensive datasets, further enhances the research collaboration of many research groups in academia, industry, and government across the nation.