面向路网环境速度预测攻击的隐私保护
A Privacy Preserving Method from Attacks of Velocity Prediction in Road Network

针对路网环境中攻击者利用速度预测获得用户位置隐私的问题,提出了一种提高当前路段查询密度值的密度压缩算法。该算法在用户真实位置附近添加大量噪声用户,通过噪声用户影响当前路段查询密度,进而降低速度预测的准确性,破坏攻击者通过概率转移矩阵预测用户行驶速度的攻击行为,以此保护用户在路网环境中的位置轨迹隐私。该算法通过密度压缩使真实用户和噪声用户表现出相同速度,提高了真实用户与噪声用户之间的相似程度,降低了噪声用户被识别的机率,进一步隐藏了真实用户。实验结果表明,与其他主流算法相比,密度压缩算法能够更有效地抵抗基于速度预测的攻击行为,具有更好的隐私保护能力。在执行时间和隐私保护成功率等方面的实验结果进一步表明,该算法更适合在路网环境下提供隐私保护服务,具有广阔的应用前景。
A density compression algorithm (DCA) to query density of current road segment is proposed to solve the problem of filching location privacy by velocity prediction. The algorithm produces a large number of dummy users and utilizes these users to increase the query density of current road segment and to decrease the prediction accuracy so that velocity prediction attacks based on the matrix of transition probability are destroyed. Thus the aim of location trajectory privacy preservation is realized. Moreover, the DCA also makes dummy users have the same velocity with the real user, which increases the similarity of dummy users with the real user and reduces the probability of dummy users to be identified so that the real user is further hidden. Experimental results and comparisons with other schemes show that the DCA resists the attacks from velocity prediction and achieves a better privacy preservation. In addition, the DCA also has a shorter running time as well as a higher success probability in privacy preservation. These characteristics show that the DCA may have a better utilization in location privacy of road networks and a more broad application prospect