| 研究課題/領域番号 |
18H06471
|
|---|
| 配分区分 | 補助金 |
|---|
| 研究機関 | 国立情報学研究所 |
|---|
研究代表者 |
Chiang YiHan 国立情報学研究所, アーキテクチャ科学研究系, 特任助教 (10824196)
|
|---|
| 研究期間 (年度) |
2018-08-24 – 2020-03-31
|
| キーワード | cotask / mixed integer program / mobile edge computing / task offloading / task scheduling |
|---|
| 研究実績の概要 |
In this research, the applicant investigates the problem of cotask-aware offloading and scheduling in mobile edge computing (MEC) systems: Cool-Edge. Mathematically, the Cool-Edge problem can be formulated as a mixed-integer non-linear program (MINLP) to minimize average cotask completion time (ACCT). To cope with the intractability of the Cool-Edge problem, the applicant has proposed the cotask-aware offloading and scheduling algorithms based on an LP rounding (LPR) technique and an earliest-cotask-arrival-first (ECAF) rule, respectively, and proved the achieved approximation factor. In addition, the proposed solution has been evaluated through both WiFi testbed experiments and simulations. The above results have been submitted in part to two IEEE journals and one IEEE conference.
|
| 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
Even though the proposed solution can reduce ACCT effectively as cotasks can be offloaded in a balanced way and scheduled synchronously, its practicability can be enhanced by addressing the following issues. (1) Adequately sequencing the transmission of multiple sub-tasks (destined from a mobile device (MD) to an edge server (ES)) can allow ESs to begin processing early. (2) Enabling preemptive processing (i.e., an ongoing job can be interrupted and resumed later) can improve the efficiency of cotask processing. (3) Designing an online competitive solution can make MEC robust to the dynamics of cotask generation, user mobility and network topologies. (4) Setting up a 4G/5G testbed is more practical than WiFi as MEC systems are very likely to be part of next-generation cellular systems.
|
| 今後の研究の推進方策 |
In the next fiscal year, the applicant aims to overcome the above challenges to add merits to this research. (1) Reformulate the Cool-Edge problem to be aware of transmission sequencing, and show there exists an optimal strategy by proving a greedy-choice property and an optimal sub-structure. (2) Redefine CCT since the preemption can break the processing of a sub-task into disjoint pieces, and refine the corresponding optimizers and constraints. (3) Extend the cotask offloading and scheduling algorithms to online ones, and prove the achieved competitive factor. (4) Implement the designed solutions over the platforms of 4G-based SINET-WADCI and NTT DoCoMo 5G Pre-service, and meanwhile introduce signaling flows to exchange control messages, thereby facilitating practical system operations.
|
