一個實現車載網路GeoBroadcast功能的軟體定義網路架構

Abstract

在此篇論文中我們實作車載網路GeoBroadcast功能的軟體定義網路架構，在傳統IP網路中，路側單元收到的所有警告訊息皆會轉送給控制中心，控制中心必須處理所有警告訊息，並且將訊息轉送至位於目的地內的路側單元，路側單元在廣播此訊息，因此造成控制中心龐大的負擔，並且消耗大量的網路頻寬。在SDN架構之中，我們實現了一個應用程式能夠自動管理路側單元位置資訊，控制器利用實驗性訊息（Vendor）來取得路側單元的位置資訊，作為GeoBroadcast決定路由目的地之依據。當SDN控制器收到第一個來自路側單元的警告訊息時，控制器利用拓樸與地理位置資訊取得路由路徑，並且在目的地區域內受影響之路側單元加入路由規則，後續的警告訊息不需要再傳遞給控制器，只需依照路由規則將訊息傳遞至目的地即可。另外我們實作兩個應用程式，其中，路側單元視覺化應用程式可將路側單元顯示於Google map之上。而警告訊息通知應用程式可讓管理者針對特定區域發出警告訊息。最後我們在OpenNet上模擬並且評估其效能，證實一個實現車載網路GeoBroadcast功能的軟體定義網路可以大幅降低控制器的負擔、訊息延遲時間和整體網路頻寬的消耗。

In this thesis, we propose a Software Defined Network (SDN) architecture for GeoBroadcast in Vehicular Ad-hoc Networks (VANETs). In a traditional IP network, RSU receives warning messages and redirect all messages up to a control center. The control center response to process and forward every warning messages dwon to RSUs whithin the destination area for geobroadcast. Therefore, it cause a huge overhead on the control center and increase network bandwidth consumption. In our SDN architecture, we implement an application in a controller to automatically manage the locations of Roadside Units (RSUs). The controller uses a Vendor type message to get the RSUs’ geographic location, which is used as a basis for our GeoBroadcast routing. While the SDN controller received the first warning message from a RSU, it will use topology and geography information to setup the routing paths on the corresponding SDN switches to the destination RSUs. For the subsequent warning message, they can follow these paths without going through the controller. We also implement two applications, one of which is RSU Visualization APP and the other being an emergency notification APP. The first APP shows the RSUs’ locations on Google map, and the second provides an administrator to send warning messages to specific region. Finally we simulate and evaluate its effectiveness on the OpenNet. We prove that a software defined network architecture for GeoBroadcast can reduce controller overhead, message latency, and network bandwidth consumption.