智慧城市、智慧服務或智慧生活這些名詞,已然在現今社會當中風行多年,這些都將歸功於科技在軟硬體方面的技術日益進步,配合萬物聯網的概念,大量的資料急速產出,人工智慧的技術的茁壯成長,各式應用更是水漲船高。然而在這其中,需要低延遲的服務亦呼之欲出,如何有效率的利用資源,以更低的時間成本取得計算結果更顯得重要。 在本篇論文以命名資料網路(NDN : Named Data Network)替代傳統網路層(Network Layer) 的TCP/IP 路由,透過NDN將封包命名的特性與以其名稱前綴(Name Prefix)作為決策路由演算法的判斷依據為基礎,應用於服務與資料的發現機制,並以此設計出一套可以應用於節點具有移動性與無線特性的路由演算法,在連線環境方面,本論文以行動隨意網路(MANET : Mobile Ad-hoc Network)為例,以此從另一層面降低NDN需倚賴路由表才能應用於移動節點的侷限性與實現NDN在非有線環境下的可行性。本研究在應用情境方面,提出一個基於邊緣運算( Edge Computing )的智慧商圈來證明路由演算法的應用性。其概念為允許商家與消費者從平台下載應用程式即可自動配置,商家除了提供網路路由,亦為服務提供者,消費者輔以網路的路由節點與資料提供者。 綜合上述,本研究聚焦於降低NDN網路在維護路由表所需要的成本、可運行於具有節點移動性與無線的環境,並且以智慧商圈來說明此演算法可以建立一個以低成本快速部屬智慧商圈。
Due to the advance of technology in software and hardware, the terms, such as smart city, smart service or smart life are popular these years. With the concept of the Internet of Everything, many devices are connected to Internet. This leads to the rapid growth of data producing. Besides, amount of data computing techniques, such as Artificial Intelligence (AI), or Virtual Reality (VR) has been utilized in many different applications. However, while using these services, it should be considered how to efficiently obtain enough appropriate data in order to achieve satisfactory computational results. In the thesis, we assume that data privacy is concerned and data is distributed saved on IoT devices. For this reason, Named Data Network (NDN: Named Data Network) is adopted to replace the TCP/IP routing algorithm. In NDN, the routing factor is based on packet naming and the name prefix, which can provide the mechanism for data and service discovery. Furthermore, NDN allows router nodes to store data as a cache for fast response. We modify the original NDN routing and design a set of new routing algorithms which can be applied in wireless environments and support node mobility. Therefore, we enhance the feasibility for NDN over non-wired connection environment. We also modify the data collection method which fit the computing purpose and low latency goal. To demonstrate the routing algorithm and our proposed platform, we choose a smart shopping area as an example which is based on edge computing architecture with the shop nodes. IoT data nodes are assumed to be customer mobile phones. Shop nodes provide service for customers, and customer nodes provide raw data for service computation, and both of them provide routing to share the network load. The NDN configuration can be fast deployed automatically among these nodes. Our simulations show that we can have minimized overhead of maintaining NDN routing tables, and get the service data and computing results back in low latency for mobile customer nodes in the smart shopping scenario.