| 摘要: | 行動物聯網環境乃整合有線、無線及行動通訊網路與物聯網遂形成下世代網路基礎，對於目前行動物聯網之發展，本計畫擬就異質網路整合、網路資源管理與服務品質保證等議題面向，透過以邊際運算、軟體定義網路與機器學習技術的導入，研發一具有動態網路資源管理之行動物聯網資料傳輸技術。計畫內容將專注於行動物聯網架構、資料傳輸與網路資源管理技術之研發，並以智慧城市規模為藍圖，據以規劃一系列的子議題及研究方法，以實現上述之研發目標。本計畫之執行規劃，首先將由小型家庭網路場域逐步拓展至大型智慧城市的網路規模，分析機器學習、邊際運算及軟體定義網路等應用於行動物聯網環境之特性及限制，探討高效的資料傳輸和網路資源管理的需求及問題，並據以提出解決方法。計畫內容主要包括以下五個部分，依序進行：第一部分，針對行動物聯網中位處邊緣的小型家庭網路服務，研發合適的機器學習模型，以提供即時性的網路頻寬分配服務；第二部分，研發適用於類社區型之多個家庭網路下動態拓樸管理及網路頻寬分配機制，力求提高區域性網路資源利用和數據傳輸量；第三部分，研發具服務品質導向之網路頻寬分配和裝置管理的機制，協助終端裝置於有限網路資源的條件之下進行最佳化的資料傳輸決策；第四部分，研發基於邊際運算和軟體定義網路之網路架構及路由最佳化，導入上述三個部分之研發機制，整合形成一套可支援智慧城市規模之物聯網架構及路由管理機制；第五部份，研發基於邊際運算和軟體定義網路之行動裝置換手與資源分配策略，有效的分流負載讓行動終端節點獲得最佳的網路資源配置以維持服務品質；本計畫執行至最後，我們將實現行動物聯網環境下具智慧化之資料傳播及網路資源管理技術，並完成上述的各機制的設計及其效能的量測和分析。 ;Mobile Internet of Things (MIoT) environments integrate wired, wireless and mobile networks, and the Internet of Things, which forms a developing basis for the next generation network. For the prospective MIoT, our study plans to investigate jointly effects on heterogeneous network integration, network resource management and service quality guarantee, and then employ edge computing, software-defined networking and machine learning technologies for developing data dissemination and network resource management mechanisms in the MIoT. Our study will focus on the development of an MIoT architecture, data dissemination and network resource management technologies. With the smart city as a blueprint, this study proposes a series of research topics and issues, and will provide the solutions to realize data dissemination with dynamic network resource adjustment in MIoT environments.Our research project will firstly analyze the characteristics and limitations of machine learning, edge computing and software-defined networks upon building an MIoT system with respect to an incremental network size from a small-scaled home network to large-scaled smart city. Meanwhile, this research project will explore potential needs and problems for efficient data dissemination and network resource management, as well as propose possible solutions to this end. Therefore, this project proposal includes five parts, as follows. First, we will develop a suitable machine learning model that can be referred to provide instant network bandwidth allocation for small home networks at the edges of the MIoT. Second, we will develop dynamic topology management and network bandwidth allocation mechanisms for multi-home networks, so as to improve network resource utilization and data throughput in community fields. Third, we will design QoS-oriented network bandwidth allocation and device management mechanisms, so as to assist terminal devices in optimizing decisions on data delivery under limited network resources. Fourth, we will design an MIoT architecture with route optimization based on edge computing and software-defined networking technologies. By incorporating the efforts of the above three parts into this architecture design, we will provide an optimal routing management mechanism to support network services in smart city scale. Finally, we will further devise a new cooperative strategy for mobile device handoff and resource allocation in this proposed MIoT architecture. This strategy will offer cooperative offloading and load-balancing services in edge areas, so that terminal devices will obtain appropriate network resources to maintain QoS for media services. In accordance with all above, our study will achieve smart data dissemination and network resource management mechanisms in MIoT environments. |
