Biblio

Website phishing continues to persist as one of the most important security threats of the modern Internet era. A major concern has been that machine learning based approaches, which have been the cornerstones of deployed phishing detection solutions, have not been able to adapt to the evolving nature of the phishing attacks. To create updated machine learning models, the collection of a sufficient corpus of real-time phishing data has always been a challenging problem as most phishing websites are short-lived. In this work, for the first time, we address these important concerns and describe an adaptive phishing detection solution that is able to adapt to changes in phishing attacks. Our solution has two major contributions. First, our solution allows for multiple organizations to collaborate in a privacy preserving manner and generate a robust machine learning model for phishing detection. Second, our solution is designed to be flexible in order to adapt to the novel phishing features introduced by attackers. Our solution not only allows for incorporating novel features into the existing machine learning model, but also can help, to a certain extent, the “unlearning” of existing features that have become obsolete in current phishing attacks. We evaluated our approach on a large real-world data collected over a period of six months. Our results achieve a high true positive rate of 97 %, which is on par with existing state-of-the art centralized solutions. Importantly, our results demonstrate that, a machine learning model can incorporate new features while selectively “unlearning” the older obsolete features.

With an increasing number of adversarial attacks against Industrial Control Systems (ICS) networks, enhancing the security of such systems is invaluable. Although attack prevention strategies are often in place, protecting against all attacks, especially zero-day attacks, is becoming impossible. Intrusion Detection Systems (IDS) are needed to detect such attacks promptly. Machine learning-based detection systems, especially deep learning algorithms, have shown promising results and outperformed other approaches. In this paper, we study the efficacy of a deep learning approach, namely, Multi Layer Perceptron (MLP), in detecting abnormal behaviors in ICS network traffic. We focus on very common reconnaissance attacks in ICS networks. In such attacks, the adversary focuses on gathering information about the targeted network. To evaluate our approach, we compare MLP with isolation Forest (i Forest), a statistical machine learning approach. Our proposed deep learning approach achieves an accuracy of more than 99% while i Forest achieves only 75%. This helps to reinforce the promise of using deep learning techniques for anomaly detection.

Internet of Thing (IoT) devices are being widely used in smart homes and organizations. An IoT device has some intended purposes, but may also have hidden functionalities. Typically, the device is installed in a home or an organization and the network traffic associated with the device is captured and analyzed to infer high-level functionality to the extent possible. However, such analysis is dynamic in nature, and requires the installation of the device and access to network data which is often hard to get for privacy and confidentiality reasons. We propose an alternative static approach which can infer the functionality of a device from vendor materials using Natural Language Processing (NLP) techniques. Information about IoT device functionality can be used in various applications, one of which is ensuring security in a smart home. We demonstrate how security policies associated with device functionality in a smart home can be formally represented using the NIST Next Generation Access Control (NGAC) model and automatically analyzed using Alloy, which is a formal verification tool. This will provide assurance to the consumer that these devices will be compliant to the home or organizational policy even before they have been purchased.

This work takes a novel approach to classifying the behavior of devices by exploiting the single-purpose nature of IoT devices and analyzing the complexity and variance of their network traffic. We develop a formalized measurement of complexity for IoT devices, and use this measurement to precisely tune an anomaly detection algorithm for each device. We postulate that IoT devices with low complexity lead to a high confidence in their behavioral model and have a correspondingly more precise decision boundary on their predicted behavior. Conversely, complex general purpose devices have lower confidence and a more generalized decision boundary. We show that there is a positive correlation to our complexity measure and the number of outliers found by an anomaly detection algorithm. By tuning this decision boundary based on device complexity we are able to build a behavioral framework for each device that reduces false positive outliers. Finally, we propose an architecture that can use this tuned behavioral model to rank each flow on the network and calculate a trust score ranking of all traffic to and from a device which allows the network to autonomously make access control decisions on a per-flow basis.

Phishing websites remain a persistent security threat. Thus far, machine learning approaches appear to have the best potential as defenses. But, there are two main concerns with existing machine learning approaches for phishing detection. The first is the large number of training features used and the lack of validating arguments for these feature choices. The second concern is the type of datasets used in the literature that are inadvertently biased with respect to the features based on the website URL or content. To address these concerns, we put forward the intuition that the domain name of phishing websites is the tell-tale sign of phishing and holds the key to successful phishing detection. Accordingly, we design features that model the relationships, visual as well as statistical, of the domain name to the key elements of a phishing website, which are used to snare the end-users. The main value of our feature design is that, to bypass detection, an attacker will find it very difficult to tamper with the visual content of the phishing website without arousing the suspicion of the end user. Our feature set ensures that there is minimal or no bias with respect to a dataset. Our learning model trains with only seven features and achieves a true positive rate of 98% and a classification accuracy of 97%, on sample dataset. Compared to the state-of-the-art work, our per data instance classification is 4 times faster for legitimate websites and 10 times faster for phishing websites. Importantly, we demonstrate the shortcomings of using features based on URLs as they are likely to be biased towards specific datasets. We show the robustness of our learning algorithm by testing on unknown live phishing URLs and achieve a high detection accuracy of \$99.7%\$.

Fast Health Interoperability Services (FHIR) is the most recent in the line of standards for healthcare resources. FHIR represents different types of medical artifacts as resources and also provides recommendations for their authorized disclosure using web-based protocols including O-Auth and OpenId Connect and also defines security labels. In most cases, Role Based Access Control (RBAC) is used to secure access to FHIR resources. We provide an alternative approach based on Attribute Based Access Control (ABAC) that allows attributes of subjects and objects to take part in authorization decision. Our system allows various stakeholders to define policies governing the release of healthcare data. It also authenticates the end user requesting access. Our system acts as a middle-layer between the end-user and the FHIR server. Our system provides efficient release of individual and batch resources both during normal operations and also during emergencies. We also provide an implementation that demonstrates the feasibility of our approach.