| 研究課題/領域番号 |
21F21376
|
|---|
| 研究機関 | 早稲田大学 |
|---|
研究代表者 |
木村 晋二 早稲田大学, 理工学術院(情報生産システム研究科・センター), 教授 (20183303)
|
|---|
| 研究分担者 |
MEYER MICHAEL 早稲田大学, 理工学術院, 外国人特別研究員
|
|---|
| 研究期間 (年度) |
2021-09-28 – 2023-03-31
|
| キーワード | photonic network on chip / Distributed Computing / Reliability of NoC / Thermal Variation / Satellites |
|---|
| 研究実績の概要 |
Over the past 5 months, the Photonic Networks-on-Chip has been studied especially on reliable routing algorithm. Photonic Networks-on-Chip are inherently more resilient to alpha particles because of using photons for communication but suffer from other forms of faults such as thermal variation. In order to control faults by the thermal variation, the microring resonators are fabricated with a flattened coil that can heat up the microring resonators. The strain-based calculation can be improved by improving both sub aspects of the algorithm. Instead of using a threshold of failed microring resonators, a performance factor is calculated based on every possible combinations of failed microring resonators in the individual switch. This way, there is no guessing whether a message can make it through the switch if two microring resonators in the same location have failed. The second point of improvement is to skip the power-based temperature estimation and replace it by separating the network into segments of single nodes and measuring the temperature of each segment. The resulting publication is almost ready for submission.In the future, the control method might be applied to advanced network such as satellite networks.
|
| 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
The research achievement is satisfactory with respect to the initial plan.There is another year for advancing the current research.
|
| 今後の研究の推進方策 |
It is necessary to investigate various materials for photonic circuits. Various materials at different thicknesses should offer different optical properties for the photonic signals. As the application of photonic network communication, Satellite Network Performance optimization are also important. LEO Satellite network performance is heavily dependent on the number of satellites that are currently within the network. Unfortunately, the cost of adding more satellites to the network is extremely expensive. In order to improve the performance per satellite, optimizations are necessary. A few possible areas are investigated such as the connection possibilities as well as servicing points, and satellite deployments for optimization candidates. The optimizations of them could lead to improved performance in these satellite networks with negligible cost differences.
|
