整合道路交通模擬器和網路模擬器：建立一個真實車輛網路模擬

Abstract

在近年來許多車輛隨意網路(VANETs)之研究中，主要的應用為資訊傳播和交通安全服務。然而，許多網路協定(network protocol)和系統設計相關的VANETs的研究中，效能的評估通常沒有使用真實的交通流資料，因此，由這些研究衍生出來的設計，實際上在現實世界中可能無法獲得良好的效能。VANETs的節點(即車輛)具有許多特性：它們擁有高移動性、網路拓樸結構往往迅速變化、車輛運動需要按照道路交通規則和遵守交通標誌。因此，交通移動模式的真實性對VANETs模擬的真實性有顯著的影響。 本論文之研究目的為提出一個整合模擬架構，以供研究VANETs車間(Inter-Vehicle Communication, IVC)通訊時使用。此一架構整合交通移動模擬器與網路模擬器。交通移動模擬器產生真實的交通移動資訊：如車輛的位置和速度，再將此資訊即時回饋給網路模擬器。以此方式，網路模擬器能夠更真實地模擬車輛移動時之通訊。具有真實的交通流資訊，能夠讓研究人員在VANETs網路協定、系統和應用程式的研究和設計更接近現實。在本研究中，我們實作城市交通模擬(Simulation of Urban Mobility, SUMO)[2]（如交通移動模擬器）和OPNET Modeler[12]（網路模擬器）的整合，以實現此一架構。 在本論文的第二部份，我們利用所開發的架構，探討網路攝影機應用的可行性。在此應用中，裝有攝影機的車輛，可以利用車間通訊，將所拍攝到的即時影像傳送給鄰近的車輛，提供鄰近車輛駕駛觀察視頻，並作出相關的反應。此一應用在駕駛者的視線被阻擋時特別有用，功能類似於安裝在路邊的反射鏡，但幾乎可以在任何地點使用。我們使用本研究開發的模擬工具，於城市場景中有交通標誌之路口、城市場景中沒有交通標誌之巷道和高速公路的三種情境中，模擬此應用，並調整不同的參數：如車輛密度、連結資料傳輸速率、影像解析度等，在不同的設置下比較系統的效能，對封包傳輸率和端到端延遲等性能指標，進行測量和比較。結果顯示，此一應用的服務品質，深受車輛密度的影響，因而容易由周邊的道路結構和交通標誌狀態所改變。此外，在大部份的狀況下，系統需要使用較高的36 Mbps連結傳輸速率，以達到駕駛者可以接受的影像品質。

Recently, the applications of Vehicular Ad-hoc Networks (VANETs) in many existing researches have mainly been related to information distribution and traffic safety. However, in the researches that aim at designing the network protocols and systems for VANETs, performance evaluation usually does not utilize realistic traffic flow data; therefore, the derived design might actually not perform well in the real world. In VANETs, the nodes, i.e., the vehicles, have many unique features: they have high mobility, and the topology often changes rapidly; the movements of the vehicles need to follow the general traffic rules and obey the traffic signs. As a result, how realistic the traffic mobility model is can have significant impact on whether the simulations of VANETs are sufficiently realistic. The main goal of this thesis is to build an integrated simulation framework, which can be utilized to study the Inter-Vehicle Communications (IVC) in VANETs. The framework integrates a traffic mobility simulator and a network simulator; in the framework, the traffic mobility simulator generates realistic traffic mobility information, such as the location and the velocity of the vehicles, and feeds it to the network simulator in real-time. The network simulator would then be able to model the communications between moving vehicles in a more realistic way. This enables the researchers to study and design the network protocols, systems, and applications in VANETs in a setting closer to the reality – with realistic traffic flow information. In this research, we implement the integration of Simulations of Urban Mobility (SUMO) (as the traffic mobility simulator) and OPNET Modeler (as the network simulator) to realize this framework. In the second part of this thesis, we utilize the developed framework to investigate the feasibility of the networked camera application. In this application, cars with cameras can utilize the IVC to send the captured video to neighboring vehicles, whose drivers can then observe the video and react accordingly. This is especially useful when the view of the driver is blocked; the functionality is similar to the mirror installed at the side of the road, but could serve the purpose at almost any location. We simulate this application in 3 different scenarios with the developed framework: intersections with traffic signals, intersections without traffic signals in the urban area, and a highway scenario, and adjusted various parameters, such as vehicle density, link data rate, and video resolution, to compare the system performance in different settings. Performance metrics such as packet delivery ratio and end-to-end delay are measured and compared. The results show that the quality of service of the tested application is strongly affected by the vehicle density, which is determined by the neighboring road structure and the traffic light status. We also found that in most applicable scenarios, it is required to use a link data rate as high as 36 Mbps so that the video quality is acceptable for the driver.