隨著即將進入 5G 的時代，物聯網即將成為下一個會大幅度改變社 會的技術，並將會有巨大的數據流量需要被處理。然而，雲端計算的 服務是藉由將終端設備所需要計算的數據集中傳輸至雲端計算，再將 其傳送回給終端設備。而當大量的數據流量出現時，集中式的雲端計 算無法滿足某些具有特定需求的終端設備，舉例來說，在車聯網技術 下需要低延遲的服務，以即時的面對突發狀況；或者是如何在智慧城 市上實現公共安全的即時維護等等。因此，邊緣運算即能補足雲端計 算這方面的空缺，邊緣運算將計算資源靠近終端設備，使其能夠快速 處理終端設備的數據以達成低延遲的結果，讓整個物聯網系統的架構 更為完善。為此，本篇論文即是研究利用 K-means 配合終端設備位置 將邊緣計算資源布建於地理環境之中，接著，使用模糊推論的方法將 終端設備的數據有效率的分配給周圍的邊緣伺服器，使整個環境的延 遲能夠優化許多。最後，再加上換手機制，以處理邊緣伺服器負載過 重的狀況以及行動終端設備在移動的過程中如何換手邊緣伺服器以達 到優化延遲的結果。;As the 5G era is coming, IoT is about to become the next technology that will greatly change the society, and there will be a great amount of data flow that needs to be processed. However, cloud computing service is to centrally transmit the data that terminal device required to the cloud server, and then transmit it back to the terminal device. When a large amount of data flows is being transmitted, centralized cloud computing cannot meet certain terminal devices with specific needs. For example, Internet of Vehicles(IoV) requires low-latency services to deal with emergencies in real time; or the smart city also needs low-latency services to achieve real-time public safety maintenance. Therefore, edge computing can solve this problem of cloud computing. Edge computing brings computing resources closer to terminal devices, enabling the IoT architecture to quickly process data, and to be more completed by processing the data with low latency. In this thesis, we use the K-means based on the location of terminal equipment to deploy edge computing resources in the geographical environment. Then, we use fuzzy inference system to efficiently offload the data of the terminal device to the nearby edge servers, which improves the latency of terminal equipment. At last, we add handoff system to deal with the problem of the overloaded edge servers, and to improve the latency of the mobile device.
