Though representing a promising approach for personalization, targeting, and recommendation, aggregation of user profiles from multiple social networks will inevitably incur a serious privacy leakage issue. In this paper, we propose a Novel Heterogeneous De-anonymization Scheme (NHDS) aiming at de-anonymizing heterogeneous social networks. NHDS first leverages the network graph structure to significantly reduce the size of candidate set, then exploits user profile information to identify the correct mapping users with a high confidence. Performance evaluation on real-world social network datasets shows that NHDS significantly outperforms the prior schemes. Finally, we perform an empirical study on privacy leakage arising from cross-network aggregation based on four real-world social network datasets. Our findings show that 39.9 percent more information is disclosed through de-anonymization and the de-anonymized ratio is 84 percent. The detailed privacy leakage of user demographics and interests is also examined, which demonstrates the practicality of the identified privacy leakage issue.

Autonomous vehicles are expected to be a disruptive technology that has the potential to revolutionize the human mobility. However, the recent research progress on intra-vehicle network (e.g., the revealing of a series of security vulnerabilities of CAN design) has demonstrated that the security issue still represents one of the major challenges of future self-driving cars. In this study, we propose a novel edge based anomaly detection system, coined VeAnDe, which exploits edge based sensor data fusion to identify the anomaly events. VeAnDe analyzes pair-wise correlations between different intra-vehicle sensors, and utilizes these correlations to examine whether an anomaly has occurred within the vehicle. More specifically, the pair-wise correlations are organized as ring architecture to reduce the computation overhead. Furthermore, the major components of VeAnDe are embedded in edge computing devices, which enables VeAnDe to be more efficient and privacy-preserving. We evaluate the performance of VeAnDe under different scenarios, and our experimental results demonstrate its feasibility and efficiency.

Along with the popularity of mobile social networks (MSNs) is the increasing danger of privacy breaches due to user location exposures. In this work, we take an initial step towards quantifying location privacy leakage from MSNs by matching the users’ shared locations with their real mobility traces. We conduct a three-week real-world experiment with 30 participants and discover that both direct location sharing (e.g., Weibo or Renren) and indirect location sharing (e.g., Wechat or Skout) can reveal a small percentage of users’ real points of interests (POIs). We further propose a novel attack to allow an external adversary to infer the demographics (e.g., age, gender, education) after observing users’ exposed location profiles. We implement such an attack in a large real-world dataset involving 22,843 mobile users. The experimental results show that the attacker can effectively predict demographic attributes about users with some shared locations. To resist such attacks, we propose SmartMask, a context-based system-level privacy protection solution, designed to automatically learn users’ privacy preferences under different contexts and provide a transparent privacy control for MSN users. The effectiveness and efficiency of SmartMask have been well validated by extensive experiments.

Privacy inference through meta-data (e.g., IP, Host) analysis of Wi-Fi traffic poses a potentially more serious threat to user privacy. First, it provides a more efficient and scalable approach to infer users' sensitive information without checking the content of Wi-Fi traffic. Second, meta-data based demographics inference can work on both unencrypted and encrypted traffic (e.g., HTTPS traffic). In this study, we present a novel approach to infer user demographic information by exploiting the meta-data of Wi-Fi traffic. We develop an inference framework based on machine learning and evaluate its performance on a real-world dataset, which includes the Wi-Fi access of 28,158 users in five months. The framework extracts four kinds of features from real-world Wi-Fi traffic and applies a novel machine learning technique (XGBoost) to predict user demographics. Our analytical results show that, the overall accuracy of inferring gender and education level of users can be 82 and 78 percent, respectively. It is surprising to show that, even for HTTPS traffic, user demographics can still be predicted at accuracy of 69 and 76 percent, respectively, which well demonstrates the practicality of the proposed privacy inference scheme. Finally, we discuss and evaluate potential mitigation methods for such inference attacks.

The increasing utilization of Electronic Control Units (ECUs) and wireless connectivity in modern vehicles has favored the emergence of security issues. Recently, several attacks have been demonstrated against in-vehicle networks therefore drawing significant attention. This paper presents an Intrusion Detection System (IDS) based on a regression learning approach which estimates certain parameters by using correlated/redundant data. The estimated values are compared to observed ones to identify abnormal contexts that would indicate intrusion. Experiments performed with real-world vehicular data have shown that more than 90% of vehicle speed data can be precisely estimated within the error bound of 3 kph. The proposed IDS is capable of detecting and localizing attacks in real-time, which is fundamental to achieve automotive security.

The mobile advertisement (ad) network is gaining an increasing interest due to the high popularity of smart phones. Previous researches on the security issues of ad network primarily focus on the privacy, permission and malware detection while less attention has been paid to the traffic consumption issue incurred by ad network. Though it is well known that ad network plays an important role in network consumption, it represents a great challenge of giving a fine-grained classification of ad networks. Inspired by this, different from any previous researches, in this study, we take the initial step towards modeling the network consumption of Ad network in Android. We develop an automatic ad analysis platform to quantify the ad network traffic consumed by android applications (app). To achieve a fine-grained quantification, we combine two sources of network traffic. On one hand, we modify the android webview and log system in system level to capture network traffic accurately. On the other hand, we capture network traffic in router level to collect detailed information of traffic packets, such as packet size and URI. We have evaluated the developed system in terms of normal apps, repacked apps and malicious apps based on the real-world dataset, which is comprised of 93 Android apps. We find out that ad traffic takes major percentage of the whole network traffic caused by Android app. We have also studied the ad library mechanism for 10 popular ad libraries. We found ads from some ad libraries use much more network traffic because they have to be fetched from remote ad libraries each time they are shown to users while other ad libraries allow apps to store ads locally.

Although privacy leaking through content analysis of Wi-Fi traffic has received an increased attention, privacy inference through meta-data (e.g. IP, Host) analysis of Wi-Fi traffic represents a potentially more serious threat to user privacy. Firstly, it represents a more efficient and scalable approach to infer users' sensitive information without checking the content of Wi-Fi traffic. Secondly, meta-data based demographics inference can work on both unencrypted and encrypted traffic (e.g., HTTPS traffic). In this study, we present a novel approach to infer user demographic information by exploiting the meta-data of Wi-Fi traffic. We develop a proof-of-concept prototype, Demographic Information Predictor (DIP) system, and evaluate its performance on a real-world dataset, which includes the Wi-Fi access of 28,158 users in 5 months. DIP extracts four kinds of features from real-world Wi-Fi traffic and proposes a novel machine learning based inference technique to predict user demographics. Our analytical results show that, for unencrypted traffic, DIP can predict gender and education level of users with an accuracy of 78% and 74% respectively. It is surprising to show that, even for HTTPS traffic, user demographics can still be predicted at a precision of 67% and 72% respectively, which well demonstrates the practicality of the proposed privacy inference scheme.

The mobile users are facing a serious risk of losing location privacy (e.g., users' location information transmitted by open advertisement network, and the reported event of involuntary tracking of mobile users in popular mobile social apps). In this study, we design and implement LocMask, a system-level solution that provides location privacy protection in Android system. LocMask achieves the tradeoff of the privacy and the utility of location based services by providing the Quality of Protection (QoP) on demand, which sets different privacy protection levels to different locations based on how sensitive these locations are. Motivated by the fact that Top locations (e.g, user's home or office) are more sensitive than less visiting locations, LocMask provides location profile management module that records the user's mobility history and ranks the locations in terms of the user's visiting frequency. With users' location profiles, LocMask can automatically determines the sensitiveness of these locations as well as their corresponding privacy protection level. LocMask is also designed to incorporate various obfuscation techniques. The effectiveness of LocMask is supported by extensive real-world data based evaluations.