| 研究課題/領域番号 |
21K17733
|
|---|
| 研究種目 |
若手研究
|
| 配分区分 | 基金 |
|---|
| 審査区分 |
小区分60060:情報ネットワーク関連
|
|---|
| 研究機関 | 京都大学 |
|---|
研究代表者 |
HE FUJUN 京都大学, 情報学研究科, 特定研究員 (90899634)
|
|---|
| 研究期間 (年度) |
2021-04-01 – 2022-03-31
|
| 研究課題ステータス |
完了 (2021年度)
|
| 配分額 *注記 |
4,290千円 (直接経費: 3,300千円、間接経費: 990千円)
2023年度: 1,430千円 (直接経費: 1,100千円、間接経費: 330千円)
2022年度: 1,430千円 (直接経費: 1,100千円、間接経費: 330千円)
2021年度: 1,430千円 (直接経費: 1,100千円、間接経費: 330千円)
|
|---|
| キーワード | robust optimization / resource allocation / NFV / queueing theory / cloud computing / probabilistic protection / survivability |
|---|
| 研究開始時の研究の概要 |
This research studies resource allocation in cloud computing systems. Robust optimization is adopted against uncertainty in failure patterns and traffic demands in clouds. This work understands and handles different pratical problems through modeling, algorithm design, and demonstration.
|
| 研究実績の概要 |
Three articles (one as the 1st author) have been accepted by high-level conferences; one article (1st author) has been published at a top-level journal. Three journal articles have been submitted. These works focused on resource allocation problems in different applications; it includes cloud computing and network function virtualization (NFV), where network failures and traffic uncertainty typically exist, which degrade the network performance.
One work developed a backup computing and transmission resource allocation model against multiple node failures. Probabilistic protection is provided for computing resource to reduce the required computing capacity. It analyzed backup transmission resource sharing in the case of multiple failures to compute the minimum required backup transmission capacity. With our analyses, a network operator can set an appropriate degree of backup transmission resource sharing based on practical requirements. For future extensions, we plan to apply probabilistic protection for both computing and transmission resources to further reduce the required network resources.
Another work introduced a robust optimization model to handle the traffic uncertainty for service deployment in NFV. It provided different approaches to solve the deployment problem. Based on it, a network operator can develop services against traffic uncertainty in a cost-efficient way. For future work, we plan to address a more accurate model to further reduce the deployment cost introduced by conservative approximation in the current one.
|
