Location Anonymization With Considering Errors and Existence Probability
针对现有位置匿名方法忽略定位误差和用户实际存在概率的问题，提出新的隐私与效用度量及高效匿名算法，在保持数据可用性的同时降低用户属性被识别的风险。
Mobile devices that can sense their location using GPS or Wi-Fi have become extremely popular. However, many users hesitate to provide their accurate location information to unreliable third parties if it means that their identities or sensitive attribute values will be disclosed by doing so. Many approaches for anonymization, such as k-anonymity, have been proposed to tackle this issue. Existing studies for k-anonymity usually anonymize each user's location so that the anonymized area contains k or more users. Existing studies, however, do not consider location errors and the probability that each user actually exists at the anonymized area. As a result, a specific user might be identified by untrusted third parties. We propose novel privacy and utility metrics that can treat the location and an efficient algorithm to anonymize the information associated with users' locations. This is the first work that anonymizes location while considering location errors and the probability that each user is actually present at the anonymized area. By means of simulations, we have proven that our proposed method can reduce the risk of the user's attributes being identified while maintaining the utility of the anonymized data.