車載網路可用來疏通交通壅塞以及提供適地性的服務，這些服務為了要在車載網路上達到快速的訊息傳送，通常會使用到虛擬骨幹的作法。本篇論文提出一個新的以路線依從性與時間可預測性來分類車輛，並以此為基礎來建構車載網路的虛擬骨幹方法－ Classification based Virtual Backbone Construction in Vehicular Ad-hoc Networks。 此方法在推廣初期能以較少的建構成本來覆蓋大多數的車輛，使駕駛者有動機來裝設車載網路的通訊設備，快速的達到車載網路的普及。使用交通模擬工具STRAW 與網路模擬工具JiST/SWANS，載入美國人口普查局TIGER 系統的舊金山市街道資料，模擬利用幹線公車來建構虛擬骨幹，可以提供比目前其他相關研究所使用協定更好的服務效率，若搭配佈建在十字路口的路側單元，可提供更可靠與更長時間的服務。 經由模擬檢驗可知利用幹線公車來建構虛擬骨幹的方式，可以用較少的通訊設備裝置，達到較高的網路覆蓋率，較長時間可提供網路服務，節點與叢聚點平均連結時間也較長，且距離骨幹最遠的點與平均節點距離，都相較其他方法近。Vehicular Ad-hoc Networks(VANET) can be used as a mean to alleviate traffic congestion and provide location-based services. The success of these applications relies heavily on effective message transmissions. To achieve this goal, the virtual backbone is known to be a powerful tool. In this thesis, we propose a novel method to classify vehicles according to their route and time predictability, and use such classification as the basis to construct the virtual backbone for VANET. Our protocol, namely Classification-based Virtual Backbone Construction in Vehicular Ad-hoc Networks (CVBCV), aims at providing network services to more vehicles with less hardware cost at the initial stage of VANET deployment. By doing so, we hope to provide enough incentive to drivers to install IEEE 802.11p On Board Unit (OBU) at their vehicles, and can eventually help speed up the deployment of VANET. To obtain a reliable simulation results, we utilize the traffic simulation tool STRAW and network simulator JiST/SWANS, with the real San Francisco street dataset downloaded from United State TIGER system. The simulation results show that by using only a few major bus routes as the virtual backbone, we are able to provide more network coverage, longer service duration, and fewer handoffs with fewer OBUs. The best results can be had when CVBCV works with a few Road Side Units deployed at traffic-concentrated crossroads.
