| 摘要: | 在現今資訊化的時代下，雲端運算受限於網路頻寬大小以及地理位置遠近等因素，沒有辦法即時滿足特定物聯網服務的需求，例如：自駕車、智慧城市等。而邊緣運算的出現正好滿足雲端運算的不足之處，邊緣運算採用分散式架構，因此比起雲端運算，邊緣運算能優化回應服務請求的速度，減少延遲，並大量減少網路頻寬的使用量以及能源與傳輸成本上的浪費。然而，在邊緣運算場景服務越發盛行的情況下，我們缺少一個在此服務情境下發生之服務的品質衡量辦法。因此，本研究旨在針對機器網絡之間發生的服務，並以更全面的角度出發，提出一應用於邊緣運算情境下之服務品質綜效架構(Edge Computing Service Quality Framework，簡稱EdgeQual)。同時，透過此一架構所建立的五個品質構面指標─可靠性、回應性、可用性、靈活性、流動性，並結合服務品質缺口理論來衡量服務品質。在實務應用上，本研究依據所提出之架構，實作出一個分散式自走車雛型系統；並將此系統運行之服務所得出的數據結果，透過網頁的方式呈現。為了驗證本研究提出的架構與所實作出之系統的可行性，本研究亦改變實驗環境的設置來測試EdgeQual是否會因為環境的改變導致服務品質結果的變動。實驗結果證實當研究環境越不好時，服務品質的表現也會越不好。最後，針對現有的運用與研究限制進行討論，並提出未來研究的可能方向。;In the era of informatization, cloud computing suffers from the network bandwidth and the geographical distance and thus is hard to fulfill the demand of specific IoT service, such as self-driving car and smart city. Due to its features of decentralization and distribution, edge computing is believed to solve the problem of cloud computing. Compare to cloud computing, edge computing has multiple advantages, it can optimize its response time to service requests and reduce latency as well. Furthermore, it can decrease the usage of network bandwidth and the waste of power and transmission cost. However, current development of edge computing lacks a measurement of service quality under edge computing scenario is still needed. As a result, this research proposed an edge computing service quality framework, EdgeQual, and focused on the services operating on machines. EdgeQual includes five dimensions: reliability, responsiveness, availability, elasticity, and mobility. Service quality of edge computing can be evaluated with the integration of EdgeQual’s five dimensions and service quality gap theory. According to the framework proposed, we implemented a distributed smart robot car system prototype to operate a service, and collect all the data produced by the service, then presented the results on the website. Moreover, to validate the EdgeQual and the system feasibility, we change the experimental condition to test if these changes of environment will affect the service quality. The result showed that if the experimental environment in a bad condition will have a negative impact on service quality. At last, we will discuss the research restriction and the possible way of future research. |
