| 摘要: | 隨著網路融入現代生活的一部分，萬物將皆為網路中的一小成員，物聯網發展蓬勃，逐漸成為未來生活的型態，如智慧家庭、智慧農場、智慧醫療、智慧工廠、智慧電表等。這些網路環境中的每一個成員皆會產生數據，形成資料流並且占用網路資源：頻寬，由於網路頻寬並非取之不盡用之不竭，傳輸連線在每個時間單位內所能負荷的頻寬是有限度的，當這些資料流超出網路頻寬的上限，在傳輸過程中相互爭奪並產生網
路壅塞的情形，讓資料抵達時間延遲，除了使用者體驗上的不佳，嚴重時甚至造成不可挽回的後果。例如防盜系統通報延遲，錯過抓捕小偷的黃金時間; 抑或是火災意外發生時，通報系統與灑水系統發生延遲，這些都是分秒必爭的過程，為了避免這些憾事的發生，本篇利用邊緣運算的基礎，在智慧家庭網路中設置邊緣節點，能有效控管網路頻寬的運用，將資料流導入軟體定義網路架構，讓靠近邊緣節點的軟體定義網路交換機來監測邊緣節點所生成的流量，讓軟體定義網路控制器能調度不同線路的資源，將資料流根據所監測的流量來進行分流來分散風險，使整體智慧家庭的網路頻寬得以有最大效益的運用。;With the integration of the network into a part of modern life, everything could connect with the network and be a member of it. The Internet of Things (IoT) is booming,
gradually becoming the coming future of daily life; many IoT applications include smart home, smart farm, smart medical, smart factory, smart meters, etc.Each device in the network produces data, forms data streams, and consumes the network resource: bandwidth. Because the network bandwidth is exhaustible reousrce, the bandwidth that can be utlized in each unit of time is limited. When the demands of these data streams exceed the upper limit of network bandwidth capacity, the devices compete with each other in the transmission process and result in network congestion and data transmission delay. This problem causes not only poor user experiences, but even irreversible consequences. For example, the anti-theft system notification delays, so that users miss the prime time to catch thieves. Another example is that a fire accident occurs, but the notification system and sprinkler system delay. These situations must race against time. In order to avoid the occurrence of above situations, this thesis uses the basis of edge computing to effectively utilize the bandwidth in smart home. We employ the edge nodes. An edge node can send the data flows into the Software-Defined Network (SDN) architecture, and the SDN Switch nearby could monitor the traffic produced by edge nodes. The SDN controller can dispatch the resource of the other routes of the topology to spread the risk. Thus, the study efforts in this thesis is able to maximum the overall throughput from various devices in smart home space. |
