| 研究課題/領域番号 |
21K11821
|
|---|
| 研究機関 | 東北大学 |
|---|
研究代表者 |
Kiselyov Oleg 東北大学, 情報科学研究科, 助教 (50754602)
|
|---|
| 研究期間 (年度) |
2021-04-01 – 2025-03-31
|
| キーワード | stream processing / info flow processing / sensor fusion / complex event processing / DSL |
|---|
| 研究実績の概要 |
Our first goal was extending the earlier stream-processing DSL (called strymonas) to support windows: finite subranges over a stream, limiting the scope of stream operations. Windowing is prevalent in signal processing applications (including sensor data processing).
The B4 student I have been advising and I implemented several optimal windowing strategies in strymonas, and applied them to FIR filtering, as typical in Software-Defined Radio (SDR). We have implemented the FM Radio processing pipeline entirely in strymonas. Our application is faster and using remarkably less memory than a StreamIt implementation in the closest related research.
With co-authors we have been working on the comprehensive paper explaining both the theory and the implemenation of strymonas.
One aspect of strymonas (used in the above FM Radio project and elsewhere) was generating C code from a high-level description. This is a non-trivial problem with many pitfalls. I have described the pitfalls and the ways we have overcome them in a Systems description paper (now in print).
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
立てた目標を達成した。結果は、指導していたB4の学生の卒論とPPL会議で発表しました。論文の作成中。
|
| 今後の研究の推進方策 |
In 2022, the student whom I am advising (who is now M1) are planning to write up the FM Radio results for an international workshop, and also compare to GNU Radio. We will be trying to apply our windowing approach in other applications, including sample NexMark queries. We are planning to finish and submit the comprehensive paper explaining the theory and implementation of strymonas.
|
| 次年度使用額が生じた理由 |
当該事業の研究によって得られた結果を論文・会議で発表する予定です。2021年度内、参加したい適切な学会は新型コロナの影響で中止したので、2022年度での参加を考えています。
|
