- 學位論文
霧計算系統之智慧任務卸載方法
An Intelligent Decision Method for Task Offloading in Fog Computing System
本文將於2024/08/11開放下載。若您希望在開放下載時收到通知，可將文章加入收藏。
摘要
在現今移動裝置蓬勃發展的社會下，許多應用服務逐漸都轉移到移動裝置上去使用。但這樣的狀況會導致一些問題及限制，如移動裝置的硬體資源沒有比一般機器好和電池能量有限，或者無法去執行一些複雜的應用程式時，就需要透過任務卸載將一些運算需求高的部分交給更高階層的機器去操作。 然而，要如何選擇一個適當的時機將任務卸載到雲霧端、卸載到哪一個適當的雲霧端、有哪些任務需要被卸載、或要用什麼方式去卸載任務，這些都是個重要的議題，許多論文針對這四個面向已做了分析及討論。且在移動網路環境下，執行效率及能量消耗一直是決定是否將任務卸載到雲霧端的著重因素，許多論文會在這兩個因素中做權衡，而我們便是希望在執行效率及能量消耗上能達到最佳化的結果。 此篇論文將改善[1] 的缺點，只針對系統和移動裝置整體的能力，並未考量到當前任務執行的狀況。所以其論文基於[1] 的霧計算架構，去設計一個更合適的任務卸載演算法，除考慮執行時間和能量消耗和其他花費外，更針對網路吞吐量及雲節點的使用負載因素去做考量。最後我們將會藉由模擬去評估我們所設計的方法與其[1] 做比較。模擬的結果顯示，我們的方法更能符合現實情況，帶給移動裝置用戶更好運算效率和效能。
並列摘要
Nowadays, the use of mobile devices has grown rapidly across the world. More and more applications or services are transferred to mobile devices. However, it will result in some problems and limitations. For example, mobile devices’ resources and battery life are not better than normal machines or they cannot execute complex applications. In this situation, it is beneficial for mobile devices to offload computation-intensive tasks to high-level machines. While choosing a suitable time to do task offloading to the cloud/fog, choosing a better place to offload, what portion of the application should be offloading, and offloading appropriately are important issues. Many papers have already analyzed and discussed according to the four aspects. Moreover, in a mobile network system, task offloading always emphasizes execution efficiency and energy consumption and does the trade-off between the two factors. Therefore, we expect to optimize both execution performance and energy consumption. In this paper, we will improve the disadvantage of [1] which just considers the cloud severs and mobile devices’ entire capability regardless of the current task execution status. Therefore, we proposed a more preferable offloading policy based on [1]’s fog computing model and offloading policy considers energy consumption, execution time, other expenses, especially throughput and load of cloud server’s utilization. Finally, we evaluate the performance of our method through simulation compared to [1]. The result from the simulation shows that our proposed method can be more preferable for reality and bring more computation effectiveness and performance to mobile users.
參考文獻
[1] Qiliang Zhu et al. “Task offloading decision in fog computing system”. In: vol. 14. 11. IEEE, 2017, pp. 59–68.
[2] Huaming Wu. “Multi-objective decision-making for mobile cloud offloading: A survey”. In: vol. 6. IEEE, 2018, pp. 3962–3976.
[3] Huaming Wu, William Knottenbelt, and Katinka Wolter. “Analysis of the energy-response time tradeoff for mobile cloud offloading using combined metrics”. In: 2015 27th International Teletraffic Congress. IEEE. 2015, pp. 134–142.
[4] Erol Gelenbe and Ricardo Lent. “Energy–QoS trade-offs in mobile service selection”. In: vol. 5. 2. Multidisciplinary Digital Publishing Institute, 2013, pp. 128–139.
[5] Huaming Wu, Qiushi Wang, and Katinka Wolter. “Tradeoff between performance improvement and energy saving in mobile cloud offloading systems”. In: 2013 IEEE international conference on communications workshops (ICC). IEEE. 2013, pp. 728–732.