Protecting Privacy with Smartphones

许多人在日常生活中随身携带智能手机。这些智能手机配备了各种传感器，如GPS或音频，并可以访问大量数据，如一个人的照片，日历或联系人。一个人可能会与其他人分享这些数据。有时这种共享是不期望的，导致侵犯个人隐私。想象一下，Alice已经将她的智能手机配置为与她的家庭成员共享她的位置。然而，有一天爱丽丝改变了她的日常生活，进入一家商店为她的配偶买生日礼物。爱丽丝忘记了她正在与家人分享她的位置，惊喜被破坏了，爱丽丝的隐私被侵犯了。或者，假设Bob允许他的同事访问他的位置，并期望他们仅在有有效需求时才检索此信息，例如召开临时会议。然而，有一天鲍勃注意到他的位置不断被他的一个同事访问，没有正当的理由。因此，鲍勃感到被跟踪，他的隐私受到侵犯。一个人的智能手机可以通过感知个人的上下文和检测异常事件来避免这两种情况下的隐私侵犯，例如用户在新的位置或授权用户发送异常数量的查询。该提案中提出的论点是，今天的智能手机具有关于其所有者的足够的上下文信息，以便为所有者构建常见的行为模型，并使用该模型来检测可能导致隐私侵犯的不常见行为。** 更详细地说，我将开发隐私增强技术，使智能手机能够通过检测两种潜在的隐私侵犯来保护其所有者：第一，由于所有者使用信息共享应用程序而发生的侵犯行为，其中侵犯行为是由所有者改变她的日常生活或改变智能手机及其所有者的日常环境引发的。第二，由于所有者使用信息共享应用程序而发生的违规行为，其中应用程序的授权用户以非预期的方式访问信息，例如发出对信息的过于频繁的请求或在不寻常的时间（例如，在晚上）。** 此外，我将开发隐私增强技术，使智能手机不仅可以检测，而且可以避免潜在的隐私侵犯，从而保护其所有者。目标是让智能手机以自动化和智能化的方式避免隐私侵犯，限制智能手机所有者的参与，防止敏感信息被共享，但仍然可以共享对请求此信息的人有用的信息。例如，智能手机可以自动替换敏感信息（例如，Alice的当前位置），具有较不敏感但仍然有用的信息（例如，Alice到达Carol位置的估计时间）。智能手机及其运行用户安装的应用程序的能力在涉及移动的用户隐私时引起了很多关注。拟议研究的长期目标是证明今天的智能手机（和未来的移动的设备，如谷歌眼镜）有足够的上下文信息，以主动保护移动的用户的隐私。** 拟议的研究将对移动的平台隐私增强技术的研究做出重要贡献。这些技术将鼓励更多的人使用智能手机和智能手机应用程序。这将有利于整个智能手机生态系统（包括应用程序开发商，智能手机提供商和网络运营商），从而有利于整个经济。

Many people carry a smartphone with them throughout their daily lives. These smartphones are equipped with various sensors, such as GPS or audio, and have access to a lot of data, such as a person's photos, calendar, or contacts. A person may share some of this data with other people. Sometimes this sharing is not expected, resulting in a violation of the person's privacy. Imagine Alice has configured her smartphone to share her location with her family members. However, one day Alice changes her routine and enters a store to buy a birthday present for her spouse. Alice forgets that she is sharing her location with her family, the surprise is spoiled and Alice's privacy is violated. Alternatively, imagine Bob gives his co-workers access to his location with the expectation that they will retrieve this information only if there is a valid need, like having an impromptu meeting. However one day Bob notices that his location is continuously being accessed by one of his co-workers for no valid reason. As a consequence, Bob feels stalked and his privacy violated.**It would have been possible for a person's smartphone to avoid the privacy violations in both scenarios by sensing the person's context and detecting unusual events, such as the user being at a new location or an authorized user sending an unusual number of queries. The argument made in this proposal is that today's smartphones have sufficient context information about their owner in order to build a common behaviour model for the owner and to use this model to detect uncommon behaviour that could potentially result in privacy violations. **In more detail, I will develop privacy-enhancing technologies that allow a smartphone to protect its owner by detecting two kinds of potential privacy violations: First, violations occurring as a consequence of the owner using an information-sharing app, where the violation is triggered by the owner making a change to her daily routine or by changes to the usual context of the smartphone and its owner. Second, violations occurring as a consequence of the owner using an information-sharing app, where an authorized user of the app accesses information in unintended ways, such as issuing overly frequent requests for the information or asking for the information at unusual times (e.g., at night). **Furthermore, I will develop privacy-enhancing technologies that allow a smartphone to protect its owner by not only detecting, but also avoiding potential privacy violations. The objective is to have a smartphone avoid the privacy violation in an automated and intelligent way that limits the involvement of the smartphone owner, prevents sensitive information from being shared, but still makes possible the sharing of information useful to the person asking for this information. For example, a smartphone could automatically replace sensitive information (e.g., Alice's current location) with less sensitive, but still useful information (e.g., Alice's estimated time-of-arrival at Carol's location).**Smartphones and their ability to run user-installed apps have caused a lot of concern when it comes to the privacy of mobile users. The long-term goal of the proposed research is to demonstrate that today's smartphones (and future mobile devices, such as Google Glass) have enough context information available to them in order to pro-actively protect the privacy of mobile users. **The proposed research will make important contributions to research in privacy-enhancing technologies for mobile platforms. These technologies will encourage more people to use smartphones and smartphone apps. This will benefit the entire smartphone ecosystem (including app developers, smartphone providers, and network operators) and therefore the economy at large.