Cooperative perception is an essential and widely discussed application of connected automated vehicles. However, the authenticity of perception data is not ensured, because the vehicles cannot independently verify the event they did not see. Many methods, including trust-based (i.e., statistical) approaches and plausibility-based methods, have been proposed to determine data authenticity. However, these methods cannot verify data without a priori knowledge. In this study, a novel approach of constructing a self-proving data from the number plate of target vehicles was proposed. By regarding the pseudonym and number plate as a shared secret and letting multiple vehicles prove they know it independently, the data authenticity problem can be transformed to a cryptography problem that can be solved without trust or plausibility evaluations. Our work can be adapted to the existing works including ETSI/ISO ITS standards while maintaining backward compatibility. Analyses of common attacks and attacks specific to the proposed method reveal that most attacks can be prevented, whereas preventing some other attacks, such as collusion attacks, can be mitigated. Experiments based on realistic data set show that the rate of successful verification can achieve 70\% to 80\% at rush hours.
翻译:然而,由于车辆无法独立核实他们所看不到的事件,因此无法确保感知数据的真实性。许多方法,包括基于信任(统计)的方法和以合理性为基础的方法,都是为了确定数据的真实性。然而,这些方法无法在没有事先知情的情况下核实数据的真实性。在本研究中,提出了从目标车辆的车牌上建立自我验证数据的新办法。通过将假名和数字牌作为共享的秘密,让多部车辆证明他们独立地了解,数据的真实性问题可以转变为一个加密问题,可以不经信任或可信性评价加以解决。我们的工作可以适应现有的工作,包括ETSI/ISO ITS标准,同时保持落后的兼容性。对常见攻击和攻击的具体分析表明,大多数攻击是可以防止的,而防止其他攻击,例如串通攻击。根据现实的数据集进行的实验表明,成功的核查速度可以在高峰时间达到70-80-%。