| 研究実績の概要 |
In the past decade, research of context awareness has focused more deeply on studies of location awareness and activity recognition, since the two aspects are involved in a large number of practical applications, ranging from location-based services, medical care to social science. The recent decade, with the ubiquitous distribution of mobile device integrated with a suite of sensors including accelerometer, microphone, GPS, WiFi, camera, digital compass and gyroscope, the device can provide a variety of contexts, such as user’s physical location, activities and orientation, etc. However, despite the huge progress in this research area, there is still much more to be investigated. For instance, 1) most of existing solutions require wearing a device, then one could ask what can be done with the absence of such a device; and 2) the accuracy of RF-based indoor localization is constrained by the unpredictable characteristics of radio signals in indoor environments and hence needs to be improved.
In this dissertation, we design two novel context-aware systems that can address the problems like those described in the previous section and learn the useful human context information such as activities, locations attention and velocity. More specific, we first present a passive FM radio source based activity recognition system. Then, we propose an WiFi based indoor device-free positioning and tracking scheme using fine-grained Channel State Information (CSI) as source information.
|
