在現今網路發達的社會中，惡意者相互串聯發動的共謀攻擊是一種常見的攻擊方法。在車載網路這個新興的領域中，也必須考慮發生共謀攻擊的可能並探討其偵測方法。 在過去關於假訊息攻擊的研究中，很少有人討論到共謀攻擊。本研究從共謀的角度出發，依據過去文獻上對不同情境上的共謀攻擊討論，定義出車載網路中的共謀假訊息攻擊，然後利用簡單的循序圖分析車輛在警訊應用服務上的行為，從中發現應用服務中的系統狀態可以使用於偵測攻擊者。我們結合車載網路的地理與時間資源限制以及應用程式執行狀態，設計出基於狀態判斷之共謀偵測機制，將其應用於偵測車載網路中的共謀假訊息攻擊上。 本研究依照所提出之警訊應用服務進行模擬實驗，在環境中有共謀群體串聯發送假訊息，比較本研究與多數決機制的假訊息比例與誤判比例，本研究提出的方法可減少的假訊息與多數決機制相比約達20%，而誤判比例也少於多數決3%至5%。 In modern world of developed networks, it is a common way of attack that malicious users collude with each other. In the new developing area of vehicular ad hoc networks, we should take collusion attack into account as well. In the past researches about bogus message attack, there are few think about collusion. In this paper, we start from the thought of collusion. By defining the collusion bogus message attack in VANETs according to collusion attacks of different scenarios in the past and with using a simple way to analyze the actions of danger warning application service on vehicles via sequence diagram, we found that the system state of application system could be used in detecting attacker. We combine the restrain of graphical and time resource and the execution state of application service to design a state-based collusion attack detection mechanism. We also exploit the mechanism to detect collusion bogus message attack in VANETs. We do simulation experiments about the proposed danger warning application service. Collusion groups collude to send bogus messages in the experimental environment. The percentage of bogus message and false negative rate will be compared between majority voting mechanism and our mechanism. The results show that our proposed mechanism can successfully suppress the number of false message by 20%, and lessen the false positive rate by 3% to 5%.
