在動態路徑導航中威脅車輛對車輛行駛效率之影響

Abstract

由於車用隨意網路(Vehicular Ad hoc Network, VANET)的發展，駕駛人能夠透過車載機和路側裝置，進行交通資訊的蒐集與傳播的工作。動態路徑導航就是以車用無線網路為基礎，透過車間通訊彙集並整合路況資訊，得到最即時且動態的路況資訊。由於在VANET中各車輛處於寡勝賽局，可能出現以私利考量為優先的威脅車輛，將會對其它車輛造成極大影響。目前對威脅車輛尚未發展出完美的驗證方法，故本研究藉由在動態路徑導航系統中加入威脅車輛，觀察威脅車輛的影響並尋求解決方法。 本研究透過車間通訊蒐集並整合路況資訊，獲得最即時的訊息，進而使用路徑導航系統找出最短時間路線並往目的地前進。本研究在此基礎模型中加入了三種威脅車輛：a. 偽造路況、b. 偽造路況和時戳、以及c. 考慮惡意訊息回向。在路況更新方法方面，除了基本的最新時戳法則和加權時戳法則這兩種方法外，還加入了考慮地理相對位置的兩種路況更新方法。 研究結果顯示，考慮惡意訊息回向的威脅車輛對網路影響最大；考慮地理位置和加權時戳法則的路況更新方法能有效抵抗威脅車輛的干擾；加權時戳法則的路況更新方法多跑的距離會比較少；而考慮地理位置和最新時戳法則的路況更新方法在擁塞度方面表現最佳。

With the development of vehicular ad-hoc network (VANET), drivers can use on-board unit (OBU) to share traffic information via inter-vehicle communication (IVC) or roadside-to-vehicle communication (RVC). Through the dynamic navigation, a route-planning system based on real-time information sharing, drivers can find smooth paths to their destinations immediately. Because of the property of the minority game in VANET, the threatening vehicles might jam traffic operations on purpose for their personal benefit. Desirable faultless validation methods for detecting such threatening vehicles have not been completely developed yet, so in this study we try to find solutions by putting threatening vehicles into the dynamic navigation model. In this study we build a dynamic navigation traffic model, and then propose schemes for handling three kinds of threatening vehicle: the weight-modified scheme, the weight-modified and timestamp-modified scheme, and the evil message-reverberated scheme. In addition to the basic traffic information sharing methods, we also propose a new traffic information sharing method that takes the relative vehicle locations into consideration. Experimental results show that the evil message-reverberated scheme demonstrate much bigger influence to VANET than the other two types. Our results also show that the traffic information update method with relative vehicle location is helpful in resisting the interference of threatening vehicles; the weighted update method can avoid unnecessary movement; and the timestamp-first update method with relative vehicle location has the best performance in terms of comfortable driving.