科技水平在近幾十年快速的進步，行動裝置的問世大大改變人們的生活方式，在資訊爆炸的現代，行動裝置需處理各種大小不同的任務需求，當本地裝置無法負荷任務需求時，就會將任務透過網路傳送至運算處理能力更加強大的雲端伺服器(Cloud Server)處理，完成之後再回傳至本地裝置。雖然雲端伺服器的運算能力非常強大，但隨著物聯網以及智慧型裝置的普及，大量的任務都必須透過骨幹網路進行傳輸，由於裝置與雲端伺服器的距離較遠，導致網路壅塞以及產生過長的傳輸延遲，已無法滿足現今許多低延遲任務的需求，而行動邊緣運算(Mobile Edge Computing)的出現，用戶可以透過行動網路將任務卸載(Offload)至每個區域所屬的行動邊緣運算伺服器上處理之後再將其回傳，而行動邊緣運算伺服器因為較靠近用戶端，可以有效減少任務的傳輸時間，更加容易滿足用戶低延遲的需求。 本篇論文提出的Subtasks pre-allocation strategy integrated delay time and cost consideration將要卸載任務切割成數個獨立的子任務，之後計算每一個子任務的花費成本，運用0/1背包問題在任務最大容忍時間內，挑選出讓子任務分配至Local UE、D2D UE、MEC Server所得到花費成本最低的組合，並以平行處理的方式將子任務進行卸載。 ;In recent years, technology has advanced very rapidly. Mobile devices have changed people′s lifestyles. Mobile devices need to handle various tasks of different sizes. When the local device cannot handle the task requirements, the task will be transmitted to the computing processing through the network. The more powerful cloud server (Cloud Server) processes, and then returns to the local device after completion. Although the computing power of the cloud server is very powerful, with the popularization of the Internet of Things and smart devices, a large number of tasks must be transmitted through the backbone network. Due to the long distance between the device and the cloud server, resulting in network congestion and excessive transmission delay, it has been unable to meet the needs of many low-latency tasks. Mobile Edge Computing (Mobile Edge Computing) allows users to offload tasks through the mobile network to the Mobile Edge Computing server in each region for processing and then send them back. Being close to the user terminal can effectively reduce the transmission time of tasks and make it easier to meet the user′s low-latency requirements. The Subtasks pre-allocation strategy integrated delay time and cost consideration proposed in this paper divides the task to be offloaded into several independent subtasks, and its purpose is to process the subtasks in parallel. Then calculate the cost of each subtask, and use the 0/1 knapsack problem to select the combination with the lowest cost for assigning subtasks to Local UE, D2D UE, and MEC Server within the maximum tolerance time of the task.
