F-CODE : A data abstraction approach for Compressive Sensing in Mobile Sensing Application

概要

論文の詳細を見る
Mobile sensing is attractive area for researchers and developers in recent years. Especially, the emergence of Smartphone accelerates this condition. Although these platforms and applications are developing more and more, there are critical issue; handling huge sensor data. The computing time of many classification or detection algorithms that are commonly used in this area depends on data length. Thus, larger data makes huge overheads. We propose F-CODE: a data abstraction approach for compressive sensing. F-CODE aims to compress the data with maintaining the feature of signals and the algorithms can apply to it despite target data is still compressed. In order to achieve the purpose, we design the feature measurement matrix phi_f. We evaluate F-CODE in terms of the accuracy of feature extracting assuming specific mobile sensing application. In the best case, F-CODE achieve the accuracy by 68.3%.
一般社団法人電子情報通信学会の論文
2014-01-16