FWFeng Wang

Feng Wang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Feng Wang has a total of 38 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • An Effective Machine Learning Based Algorithm for Inferring User Activities From IoT Device Events

    by Guoliang Xue, Yinxin Wan, Xuanli Lin, Kuai Xu, Feng Wang

    The rapid and ubiquitous deployment of Internet of Things (IoT) in smart homes has created unprecedented opportunities to automatically extract environmental knowledge, awareness, and intelligence. Many existing studies have adopted either machine learning approaches or deterministic approaches to infer IoT device events and/or user activities from network traffic in smart homes. In this paper, we study the problem of inferring user activity patterns from a sequence of device events by first deterministically extracting a small number of representative user activity patterns from the sequence of device events, then applying unsupervised learning to compute an optimal subset of these user activity patterns to infer user activity patterns. Based on extensive experiments with sequences of device events triggered by 2,959 real user activities and up to 30,000 synthetic user activities, we demonstrate that our scheme is resilient to device malfunctions and transient failures/delays, and outperforms the state-of-the-art solution.


  • Open Access

    A Holistic Curriculum Towards Teaching Smart Home Security

    by Feng Wang, Kuai Xu, Guoliang Xue

    Smart homes with various Internet of Things (IoT) devices generate a large amount of network traffic carrying rich information and play an important role in our lives. However, there is a lack of educational material with real case studies to teach undergraduate students smart home security. In this poster paper, we present a holistic curriculum design consisting of five components teaching conceptual understanding of smart home vulnerabilities, capturing and interpreting network traffic for normal and abnormal behaviors of home users and smart devices, and writing reports on smart home security analysis and defense recommendations. A case study on a known vulnerability of Ring doorbell which reveals the password of the smart home Wi-Fi during the initial setup stage is illustrated with the network topology and the captured http traffic. The project is at its initial stage of development by students in a cybersecurity concentration at a primarily undergraduate institution. Students have started collecting and analyzing smart home traffic as a project in the Wireless Network and Security class or through supervised individual undergraduate research studies.


  • IoTAthena: Unveiling IoT Device Activities From Network Traffic

    by Yinxin Wan, Kuai Xu, Feng Wang, Guoliang Xue

    The recent spate of cyber attacks towards Internet of Things (IoT) devices in smart homes calls for effective techniques to understand, characterize, and unveil IoT device activities. In this paper, we present a new system, named IoTAthena, to unveil IoT device activities from raw network traffic consisting of timestamped IP packets. IoTAthena characterizes each IoT device activity using an activity signature consisting of an ordered sequence of IP packets with inter-packet time intervals. IoTAthena has two novel polynomial time algorithms, sigMatch and actExtract . For any given signature, sigMatch can capture all matches of the signature in the raw network traffic. Using sigMatch as a subfunction, actExtract can accurately unveil the sequence of various IoT device activities from the raw network traffic. Using the network traffic of heterogeneous IoT devices collected at the router of a real-world smart home testbed and a public IoT dataset, we demonstrate that IoTAthena is able to characterize and generate activity signatures of IoT device activities and accurately unveil the sequence of IoT device activities from raw network traffic.


  • Characterizing and Mining Traffic Patterns of IoT Devices in Edge Networks

    by Yinxin Wan, Kuai Xu, Feng Wang, Guoliang Xue

    As connected Internet-of-things (IoT) devices in smart homes, smart cities, and smart industries continue to grow in size and complexity, managing and securing them in distributed edge networks have become daunting but crucial tasks. The recent spate of cyber attacks exploiting the vulnerabilities and insufficient security management of IoT devices have highlighted the urgency and challenges for securing billions of IoT devices and applications. As a first step towards understanding and mitigating diverse security threats of IoT devices, this paper develops an IoT traffic measurement framework on programmable and intelligent edge routers to automatically collect incoming, outgoing, and internal network traffic of IoT devices in edge networks, and to build multidimensional behavioral profiles which characterize who, when, what, and why on the behavioral patterns of IoT devices based on continuously collected traffic data. To the best of our knowledge, this paper is the first effort to shed light on the IP-spatial, temporal, entropy, and cloud service patterns of IoT devices in edge networks, and to explore these multidimensional behavioral fingerprints for IoT device classification, anomaly traffic detection, and network security monitoring for vulnerable and resource-constrained IoT devices on the Internet.


  • Characterizing DNS Behaviors of Internet of Things in Edge Networks

    by Kuai Xu, Feng Wang, Sergio Jimenez, Andrew Lamontagne, John Cummings, Mitchell Hoikka

    The recent spate of cyber attacks and security threats towards Internet-of-things (IoT) systems in smart cities, smart homes, and industry 4.0 calls for effective techniques to understand if, when, who, what IoT systems are exploited and compromised by Internet attackers. Towards this end, this paper attempts to study DNS behavioral patterns of IoT systems in edge networks as a first step of characterizing their communication patterns and their interactions with IoT users, cloud servers, and other IoT or non-IoT devices in the same edge networks. Specifically, we analyze the temporal-spatial patterns of DNS behaviors of a variety of IoT systems in two dozens of edge networks, and develop a simple yet effective bloom filter mechanism for detecting anomalous traffic patterns based on unusual DNS queries and answers. To the best of our knowledge, this paper is the first effort to systematically measure and monitor IoT network traffic from DNS perspective for providing the security of heterogeneous IoT systems and ensuring IoT user privacy.


  • Mitigating the Impact of Data Sampling on Social Media Analysis and Mining

    by Kuai Xu, Feng Wang, Haiyan Wang, Yufang Wang, Ying Zhang

    The last decade has witnessed the explosive growth of online social media in users and contents. Due to the unprecedented scale and the cascading power of the underlying social networks, social media has created a new paradigm for sharing information, broadcasting breaking news, and reporting real-time events by any user from anywhere at any time. Many popular social media sites including Twitter provide streaming data services by standard APIs to the broad researcher and developer communities. Given the sheer data volume, rapid velocity, and feature variety of online social media, these sites often supply only a sampled set of streaming data, rather than the full data set to reduce the resource cost of computations, storage, and network bandwidth. In light of the substantial impact of sampling in Twitter data stream, this article explores a combination of spectral clustering, locality-sensitive hashing (LSH), latent Dirichlet allocation (LDA) topic modeling, and differential equation modeling to mitigate the impact of sampling on social media data analysis, in particular on detecting real-world events and predicting information diffusion. Our extensive experiments demonstrate that our proposed method is able to detect effectively the real-time emerging events and predict accurately the cascading pattern of these events from the 1% sampled Twitter data stream. To the best of our knowledge, this article is the first effort to introduce a systematic methodology to study and mitigate the impact of data sampling on social media analysis and mining.


  • Open Access

    Regional Influenza Prediction with Sampling Twitter Data and PDE Model

    by Yufang Wang, Kuai Xu, Yun Kang, Haiyan Wang, Feng Wang, Adrian Avram

    The large volume of geotagged Twitter streaming data on flu epidemics provides chances for researchers to explore, model, and predict the trends of flu cases in a timely manner. However, the explosive growth of data from social media makes data sampling a natural choice. In this paper, we develop a method for influenza prediction based on the real-time tweet data from social media, and this method ensures real-time prediction and is applicable to sampling data. Specifically, we first simulate the sampling process of flu tweets, and then develop a specific partial differential equation (PDE) model to characterize and predict the aggregated flu tweet volumes. Our PDE model incorporates the effects of flu spreading, flu recovery, and active human interventions for reducing flu. Our extensive simulation results show that this PDE model can almost eliminate the data reduction effects from the sampling process: It requires lesser historical data but achieves stronger prediction results with a relative accuracy of over 90% on the 1% sampling data. Even for the more aggressive data sampling ratios such as 0.1% and 0.01% sampling, our model is still able to achieve relative accuracies of 85% and 83%, respectively. These promising results highlight the ability of our mechanistic PDE model in predicting temporal–spatial patterns of flu trends even in the scenario of small sampling Twitter data.


  • Open Access

    On Consistency of Graph-based Semi-supervised Learning

    by Chengan Du, Yunpeng Zhao, Feng Wang

    Graph-based semi-supervised learning is one of the most popular methods in machine learning. Some of its theoretical properties such as bounds for the generalization error and the convergence of the graph Laplacian regularizer have been studied in computer science and statistics literatures. However, a fundamental statistical property, the consistency of the estimator from this method has not been proved. In this article, we study the consistency problem under a non-parametric framework. We prove the consistency of graph-based learning in the case that the estimated scores are enforced to be equal to the observed responses for the labeled data. The sample sizes of both labeled and unlabeled data are allowed to grow in this result. When the estimated scores are not required to be equal to the observed responses, a tuning parameter is used to balance the loss function and the graph Laplacian regularizer. We give a counterexample demonstrating that the estimator for this case can be inconsistent. The theoretical findings are supported by numerical studies.