- 學位論文
行動雲彈性運算架構的設計與實作
Elastic Computation Framework in Fog/Mobile Cloud Computing Environment
摘要
現今在行動運算上的運用越來越多樣化，許許多多的行動運用藉著行動裝置與雲端科技的結合改變了我們的生活方式與型態，同時我們發現，行動裝置的運算能力其實也是越來越強，是否可以利用這一點來增進整個行動運算的運算架構來補足對於網路方面的依賴？ 本篇論文從這個觀點出發，提出了一個可以利用週邊裝置運算的運算框架，使得我們即使不把資料送到雲端也可以完成複雜的計算要求， 這讓整個行動運算對於網路的要求大幅下降，只要利用週邊閒置的運算裝置就可以達成的任務。 在這之中，本篇論文也提出了在這個運算框架裡如何去作任務排程。 我們結合了非精確計算的運算模型與使用者的移動模型，藉由這個模型我們可以有效的規劃以及使用周遭提供的運算資源，使得我們可以作更有效率的運算而不浪費資源，我們也提出了一個排程演算法來針對這種運算模型作排程，藉由這個排程演算法，我們可以確實的達到有效利用資源的這個目標。 在最後實驗的部份，我們藉由模擬的方式來比較我們提出的演算法與其他已知演算法的差異，實驗結果顯示我們的演算法雖然不會完成最多的工作，但是系統資源的性價比會是最好的，這證明我們的演算法會確實讓資源得到最有效的利用。
並列摘要
The applications of mobile computing are various. More and more mobile applications change our live by the mobile device and the cloud computing technology. In the meanwhile, we found the computation power of the mobile device also become more and more powerful. From the observation, maybe we can improve the mobile computing architecture by the growth of the computation power of mobile devices and reduce the network requirement of mobile computing. This thesis starts from the observation above and proposes a framework which can leverage the computation resource of the idle device which around us. We can complete a complicated task even if we do not need the help of the cloud. It makes the network requirement of mobile computing decreasing. We can achieve the task only with the help of the device around us. In addition, we propose the scheduling problem in this thesis. We combine the imprecise computation model and user mobility. We can be more accuracy to scheduling the task and make less waste of the resources. We also propose a scheduling algorithm to solve it. By the algorithm, we can fulfill the target of using the resource efficiently. In the experiment, we compare our scheduling algorithm with other known algorithm by simulation. The result shows that our scheduling algorithm does not complete the most tasks though, the cost-performance of system resources is the best. It proves the system resources are truely used efficiently.