- 學位論文
SocialCloaking：a Distributed Architecture for K-anonymity Location Privacy Protection
SocialCloaking：a Distributed Architecture for K-anonymity Location Privacy Protection
摘要
隨著智慧型手機的普及與先進感測技術的發展，許多感應裝置被應用在智慧型手機上，使智慧型手機具有連結網路與定位的能力，基於適地性服務(Location-Based Service, LBS)的應用也隨之被廣泛應用。由於使用適地性服務的查尋會包含使用者的位置，使用服務會暴露使用者的位置，引起了隱私上的擔憂。匿名法是一項普遍被用來保護位置隱私的技術。在過去的文獻中，最常見的是集中式的架構，需要建置一台受信任的匿名伺服器。然而集中式的作法具有幾項明顯的弱點，如伺服器效能的瓶頸、單點故障的問題、容易成為攻擊的目標等。更重要的是，有些使用者可能不願信任匿名伺服器。在本文中，我們提出一種利用同儕計算(Peer-to-Peer)獨特的架構及線上社群朋友間的信任關係為基礎的分散式架構，在此架構下使用者可自行保管自己的位置資訊及決定要信任的朋友，並可達到不需要第三方信任匿名伺服器的介入即可達成位置保護的目的。我們利用實際使用者資料及線上社群關係來評估我們提出的架構與方法。此外我們也在具有小世界(Small World)特性的拓樸上進行多項評估。我們的實驗結果證實我們的實驗方法可以產生合理大小的模糊位置區域並可抵擋一些常見的攻擊。
並列摘要
With the popularity of smartphones and sensing technologies, many sensing devices are have widely equipped in smartphones. Smartphones have the ability to link to the internet and collect position information, so that location-based services (LBS) applications are also become popular. Since the query of LBS contains user’s location, it raises a privacy concern of exposure of user’s location. K-anonymity is a commonly adopted technique for location privacy protection. In the literature, a centralized architecture which consists of a trusted anonymity server is widely adopted. However, this approach exhibits several apparent weaknesses, such as single point of failure, performance bottleneck, serious security threats, and not trustable to users, etc. In this paper, we re-examine the location privacy protection problem in LBS applications. We propose a distributed architecture based on the unique structure of Peer-to-Peer systems and trust relationship among online social community friends. In this architecture, user can maintain their own location information and decide which friends to trust, and can achieve protecting location privacy without the third party trusted anonymous server. We use the reality user information and online social relationship to evaluate our proposed architecture and method. Furthermore, we experiments are also performed on artificially generated small world topologies. Our experiment results confirm our method can generate appropriate size of blur location region and can resist some common attack.