車載隨意網路(vehicular ad-Hoc network, VANET)，近來受到大家的重視，為實作智慧型運輸系統重要技術之一。運用車輛與道路設施作為節點之間的溝通來提供可用的訊息給予用路人。而訊息無非是能夠提供高價值與高可信度的資料，讓使用者能夠透過訊息就能掌握目前最新的狀況，因此倘若訊息遭受到偽造、竄改、延誤時，即會影響用路人的駕駛行為，更會進一步的造成交通之混亂。
因此在本研究中分為兩階段來作為判斷，先透過分層的方式傳送與接收車輛的燈號與相關資訊，並且確保其資料的可行性與完整性，再透過一套基於事件的聲譽系統來判定事故的可信度，並且利用可信度配合門檻值來發送警報訊息給用路人。本篇最後會利用城市交通模擬作為實驗，藉此進而取得更為真實的在高速公路上車輛行駛狀況與發生事故時車輛的反應行為，依據實驗結果說明當事故發生時，利用本系統提出之偵測燈號與搭配聲譽機制方式可達成的事故判斷準確率。
In recent years, Vehicular ad hoc networks (VANET) has been attracted more atten-tion, and It is also one of the important technologies of Intelligent Transportation system (ITS). The messages provided by vehicles and road facilities could be useful information to passengers. To provide passengers latest status of traffic, the messages which vehicles and road facilities provide should be high value and credibility. When these messages are delayed or tampered, the driving behavior of passenger will change which can cause traf-fic chaos.
In this paper, the proposed system uses two stage model to ensure the feasibility and integrity of VANET messages. In first stage, concept of multiple layer is used for trans-mission of traffic light information. In second stage, a set of event-based reputation system is used to determine the credibility of the accident. Alarm messages will be sent when the score exceed a pre-defined threshold. To evaluate the performance of proposed system an experiment is conducted to simulate accident on the highway. The result shows that the proposed system could detect accident correctly on highway.
