The MOBED(mobile embedded system) research group at Yonsei Univesity has actively been participating in the research and development of a wide range of embedded and wireless/mobile system software. We are currently interested in (1) embedded operating systems, (2) mobile sensing systems, and (3) context-aware green computing.
1. Embedded operating system
We aim to develop core techniques for target-specific power/performance-aware embedded operating systems, and their optimization. Our current interests include the development of personalized power management techniques for Linux-based handheld devices such as smartphones. We have developed a suite of software, the AppScope Suite(DevScope AppScope, AppScopeViewer), which provides a fine-grained and runtime energy information of Android application, running on diverse types of smartphone platforms. We are presently expanding the AppScope framework to enable the performance and power modeling of Multi-core CPU, GPU, AMOLED display, LTE, eMMC, and various sensor components. We are also developing a light-weight Linux-kernel module to monitor user’s energy-behavior, and developing relevant techniques for personalized energy management of smartphone user.
2. Mobile sensing system
Our ultimate goal is to understand the “human behavior” with spatio-temporal information, sensed from his/her own smartphone, as well as data collected from the crowd (CrowdSensing). As a first step toward this goal, we have developed a smartphone-based mobile sensing platform, called LifeMap, which enables the automatic detection of user’s POI (point of interest) and his/her routes in everyday life, even in indoors. The core techniques we have developed include the key iLBS (indoor location based service) components, such as inertial sensor-based PDR(pedestrian dead reckoning), automatic WiFi fingerprinting(Auto-fingerprinting), indoor map generation(SmatSLAM, MRI, FingdingMiMo), energy-aware sensing scheduling(SmartDC, FastTrack), automatic POI categorization (CSP), and many more. We are currently expanding the functionality of LifeMap, hoping to use it as a ground tool for our CrowdSensing research.
3. Context-aware green computing
As for our research on green computing (i.e., smart grid), we have particular interests in the development of context-aware home-energy management. Based on the user context acquired with mobile sensing tools, we can actively manage home appliances, hence reducing overall energy consumption. Towards this goal, we have developed a prototype hardware and software platform, which includes a SmartBox, gateway server, and smartphone-based application. The SmartBox is a hardware module that is connected the server via power-line communication(PLC) and provides the capability of monitoring and controlling up to 6 power outlets simultaneously. The power outlet can be controlled (on/off) remotely via smartphone. With this platform, we are currently developing various techniques such as supervised/unsupervised NILM (non-intrusive load monitoring), stand-by power management, intelligent appliance control, and so on.