| Project/Area Number |
17H04677
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (A)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Computer system
|
|---|
| Research Institution | Tokyo Institute of Technology |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2017-04-01 – 2020-03-31
|
| Project Status |
Completed (Fiscal Year 2020)
|
| Budget Amount *help |
¥13,910,000 (Direct Cost: ¥10,700,000、Indirect Cost: ¥3,210,000)
Fiscal Year 2019: ¥4,030,000 (Direct Cost: ¥3,100,000、Indirect Cost: ¥930,000)
Fiscal Year 2018: ¥5,850,000 (Direct Cost: ¥4,500,000、Indirect Cost: ¥1,350,000)
Fiscal Year 2017: ¥4,030,000 (Direct Cost: ¥3,100,000、Indirect Cost: ¥930,000)
|
|---|
| Keywords | 近似計算 / ハードウェア/ソフトウェア協調設計 / Internet of Things / Approximate Computing / 組み込みシステム / 組込みシステム / ソフトウェア解析 / 教師なし学習 / 計算機システム |
|---|
| Outline of Final Research Achievements |
Most applications in Internet-of-Things (IoT) are found to be able to approximate parts of their implementations without incurring noticeable output results. Traditionally, in signal processing, some approximation that is not perceptible to human senses have been employed only in seat of experts’ pants approaches. This research aimed to develop a novel design paradigm in embedded systems designs that aggressively and systematically leverages tolerable approximations of a target application. First, a framework to statically analyze approximatable parts, SSA-AC, has been developed. Furthermore, on three representative IoT applications, we demonstrated that our proposed approximation techniques can efficiently outperform conventional (precise) hardware acceleration techniques.
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究成果によって、積極的かつ体系的に近似計算を適用するための解析手法を初めて構築した。本手法はハードウェア/ソフトウェアによらず適用でき、組込みシステム全体の設計を効率化できる。また、3つのアプリケーションの実装例から、従来手法に勝る有効性を実証した。両研究成果とも社会的意義は大きい。また、初めての定性的解析モデルの構築、および、専用ハードウェア設計に勝るエネルギー高効率なアクセラレータ設計を実現したことから、学術的意義も高く評価され、一流国際論文誌から両成果を発表済みである。
|
