Processing burden reduction of a large-scale data compression by employing averaged compressed sensing (ソフトウェア無線)

概要

論文の詳細を見る
Compressed sensing has been drawn much attention during last few years as an eminent solution for ill-posed inverse problem. However, compressed sensing is inherently vulnerable for the large-scale data process due to its time-consuming reconstruction process. Therefore, this paper pursues a solution to solve the scalability problem in compressed sensing. This paper proposes an averaged compressed sensing method which reduces the signal processing burden in dealing with large-scale data as follows: First, large-scale data which is sampled over Nyquist rate is divided into multiple groups. Then, conventional measurement is conducted for each group with identical measurement matrix. Second, multiple measurements data is averaged. Alternatively, the averaging process can be conducted prior to the measurement process. Experiments verify the practical validity of the proposed method.
社団法人電子情報通信学会の論文
2011-01-20