In recent years, mobile devices are equipped with increasingly advanced sensing and computing capabilities. Coupled with advancements in Deep Learning (DL), this opens up countless possibilities for meaningful applications, e.g., for medical purposes and in vehicular networks. Traditional cloud-based Machine Learning (ML) approaches require the data to be centralized in a cloud server or data center. However, this results in critical issues related to unacceptable latency and communication inefficiency. To this end, Mobile Edge Computing (MEC) has been proposed to bring intelligence closer to the edge, where data is produced. However, conventional enabling technologies for ML at mobile edge networks still require personal data to be shared with external parties, e.g., edge servers. Recently, in light of increasingly stringent data privacy legislations and growing privacy concerns, the concept of Federated Learning (FL) has been introduced. In FL, end devices use their local data to train an ML model required by the server. The end devices then send the model updates rather than raw data to the server for aggregation. FL can serve as an enabling technology in mobile edge networks since it enables the collaborative training of an ML model and also enables DL for mobile edge network optimization. However, in a large-scale and complex mobile edge network, heterogeneous devices with varying constraints are involved. This raises challenges of communication costs, resource allocation, and privacy and security in the implementation of FL at scale. In this survey, we begin with an introduction to the background and fundamentals of FL. Then, we highlight the aforementioned challenges of FL implementation and review existing solutions. Furthermore, we present the applications of FL for mobile edge network optimization. Finally, we discuss the important challenges and future research directions in FL.

/pdf/federated-learning-in-mobile-edge-networks-a-comprehensive-5dgelc5m6h.pdf

Federated Learning in Mobile Edge Networks: A Comprehensive Survey

In recent years, mobile devices are equipped with increasingly advanced sensing and computing capabilities. Coupled with advancements in Deep Learning (DL), this opens up countless possibilities for meaningful applications. Traditional cloudbased Machine Learning (ML) approaches require the data to be centralized in a cloud server or data center. However, this results in critical issues related to unacceptable latency and communication inefficiency. To this end, Mobile Edge Computing (MEC) has been proposed to bring intelligence closer to the edge, where data is produced. However, conventional enabling technologies for ML at mobile edge networks still require personal data to be shared with external parties, e.g., edge servers. Recently, in light of increasingly stringent data privacy legislations and growing privacy concerns, the concept of Federated Learning (FL) has been introduced. In FL, end devices use their local data to train an ML model required by the server. The end devices then send the model updates rather than raw data to the server for aggregation. FL can serve as an enabling technology in mobile edge networks since it enables the collaborative training of an ML model and also enables DL for mobile edge network optimization. However, in a large-scale and complex mobile edge network, heterogeneous devices with varying constraints are involved. This raises challenges of communication costs, resource allocation, and privacy and security in the implementation of FL at scale. In this survey, we begin with an introduction to the background and fundamentals of FL. Then, we highlight the aforementioned challenges of FL implementation and review existing solutions. Furthermore, we present the applications of FL for mobile edge network optimization. Finally, we discuss the important challenges and future research directions in FL

A privacy preserving association rule mining algorithm was introducedThis algorithm preserved privacy of individual values by computing scalar productMeanwhile the algorithm of computing scalar product was given and the security was analyzed

Privacy preserving association rule mining in vertically partitioned data

With the exponential growth of cyber-physical systems (CPSs), new security challenges have emerged. Various vulnerabilities, threats, attacks, and controls have been introduced for the new generation of CPS. However, there lacks a systematic review of the CPS security literature. In particular, the heterogeneity of CPS components and the diversity of CPS systems have made it difficult to study the problem with one generalized model. In this paper, we study and systematize existing research on CPS security under a unified framework. The framework consists of three orthogonal coordinates: 1) from the security perspective, we follow the well-known taxonomy of threats, vulnerabilities, attacks and controls; 2) from the CPS components perspective, we focus on cyber, physical, and cyber-physical components; and 3) from the CPS systems perspective, we explore general CPS features as well as representative systems (e.g., smart grids, medical CPS, and smart cars). The model can be both abstract to show general interactions of components in a CPS application, and specific to capture any details when needed. By doing so, we aim to build a model that is abstract enough to be applicable to various heterogeneous CPS applications; and to gain a modular view of the tightly coupled CPS components. Such abstract decoupling makes it possible to gain a systematic understanding of CPS security, and to highlight the potential sources of attacks and ways of protection. With this intensive literature review, we attempt to summarize the state-of-the-art on CPS security, provide researchers with a comprehensive list of references, and also encourage the audience to further explore this emerging field.

https://ieeexplore.ieee.org/ielaam/6488907/8172502/7924372-aam.pdf

Cyber-Physical Systems Security—A Survey

An attack graph is a succinct representation of all paths through a system that end in a state where an intruder has successfully achieved his goal. Today Red Teams determine the vulnerability of networked systems by drawing gigantic attack graphs by hand. Constructing attack graphs by hand is tedious, error-prone, and impractical for large systems. By viewing an attack as a violation of a safety property, we can use off-the-shelf model checking technology to produce attack graphs automatically: a successful path from the intruder's viewpoint is a counterexample produced by the model checker In this paper we present an algorithm for generating attack graphs using model checking as a subroutine. Security analysts use attack graphs for detection, defense and forensics. In this paper we present a minimization analysis technique that allows analysts to decide which minimal set of security measures would guarantee the safety of the system. We provide a formal characterization of this problem: we prove that it is polynomially equivalent to the minimum hitting set problem and we present a greedy algorithm with provable bounds. We also present a reliability analysis technique that allows analysts to perform a simple cost-benefit trade-off depending on the likelihoods of attacks. By interpreting attack graphs as Markov Decision Processes we can use the value iteration algorithm to compute the probabilities of intruder success for each attack the graph.

Two Formal Analyses of Attack Graphs

Deep learning technologies, which are the key components of state-of-the-art Artificial Intelligence (AI) services, have shown great success in providing human-level capabilities for a variety of tasks, such as visual analysis, speech recognition, and natural language processing and etc. Building a production-level deep learning model is a non-trivial task, which requires a large amount of training data, powerful computing resources, and human expertises. Therefore, illegitimate reproducing, distribution, and the derivation of proprietary deep learning models can lead to copyright infringement and economic harm to model creators. Therefore, it is essential to devise a technique to protect the intellectual property of deep learning models and enable external verification of the model ownership. In this paper, we generalize the "digital watermarking'' concept from multimedia ownership verification to deep neural network (DNNs) models. We investigate three DNN-applicable watermark generation algorithms, propose a watermark implanting approach to infuse watermark into deep learning models, and design a remote verification mechanism to determine the model ownership. By extending the intrinsic generalization and memorization capabilities of deep neural networks, we enable the models to learn specially crafted watermarks at training and activate with pre-specified predictions when observing the watermark patterns at inference. We evaluate our approach with two image recognition benchmark datasets. Our framework accurately (100%) and quickly verifies the ownership of all the remotely deployed deep learning models without affecting the model accuracy for normal input data. In addition, the embedded watermarks in DNN models are robust and resilient to different counter-watermark mechanisms, such as fine-tuning, parameter pruning, and model inversion attacks.

Protecting Intellectual Property of Deep Neural Networks with Watermarking

The popularity of automatic speech recognition (ASR) systems, like Google Assistant, Cortana, brings in security concerns, as demonstrated by recent attacks. The impacts of such threats, however, are less clear, since they are either less stealthy (producing noise-like voice commands) or requiring the physical presence of an attack device (using ultrasound speakers or transducers). In this paper, we demonstrate that not only are more practical and surreptitious attacks feasible but they can even be automatically constructed. Specifically, we find that the voice commands can be stealthily embedded into songs, which, when played, can effectively control the target system through ASR without being noticed. For this purpose, we developed novel techniques that address a key technical challenge: integrating the commands into a song in a way that can be effectively recognized by ASR through the air, in the presence of background noise, while not being detected by a human listener. Our research shows that this can be done automatically against real world ASR applications. We also demonstrate that such CommanderSongs can be spread through Internet (e.g., YouTube) and radio, potentially affecting millions of ASR users. Finally we present mitigation techniques that defend existing ASR systems against such threat.

/pdf/commandersong-a-systematic-approach-for-practical-3570jhshgc.pdf

Commandersong: a systematic approach for practical adversarial voice recognition

An app market's vetting process is expected to be scalable and effective. However, today's vetting mechanisms are slow and less capable of catching new threats. In our research, we found that a more powerful solution can be found by exploiting the way Android malware is constructed and disseminated, which is typically through repackaging legitimate apps with similar malicious components. As a result, such attack payloads often stand out from those of the same repackaging origin and also show up in the apps not supposed to relate to each other.

Based upon this observation, we developed a new technique, called MassVet, for vetting apps at a massive scale, without knowing what malware looks like and how it behaves. Unlike existing detection mechanisms, which often utilize heavyweight program analysis techniques, our approach simply compares a submitted app with all those already on a market, focusing on the difference between those sharing a similar UI structure (indicating a possible repackaging relation), and the commonality among those seemingly unrelated. Once public libraries and other legitimate code reuse are removed, such diff/common program components become highly suspicious. In our research, we built this "Diff-Com" analysis on top of an efficient similarity comparison algorithm, which maps the salient features of an app's UI structure or a method's control-flow graph to a value for a fast comparison. We implemented MassVet over a stream processing engine and evaluated it nearly 1.2 million apps from 33 app markets around the world, the scale of Google Play. Our study shows that the technique can vet an app within 10 seconds at a low false detection rate. Also, it outperformed all 54 scanners in VirusTotal (NOD32, Symantec, McAfee, etc.) in terms of detection coverage, capturing over a hundred thousand malicious apps, including over 20 likely zero-day malware and those installed millions of times. A close look at these apps brings to light intriguing new observations e.g., Google's detection strategy and malware authors' countermoves that cause the mysterious disappearance and reappearance of some Google Play apps.

/pdf/finding-unknown-malice-in-10-seconds-mass-vetting-for-new-2l9lhk32ym.pdf

Finding unknown malice in 10 seconds: mass vetting for new threats at the Google-play scale

In recent years, as mobile smart device sales grow quickly, the development of mobile applications (apps) keeps accelerating, so does mobile app repackaging. Attackers can easily repackage an app under their own names or embed advertisements to earn pecuniary profits. They can also modify a popular app by inserting malicious payloads into the original app and leverage its popularity to accelerate malware propagation. In this paper, we propose ViewDroid, a user interface based approach to mobile app repackaging detection. Android apps are user interaction intensive and event dominated, and the interactions between users and apps are performed through user interface, or views. This observation inspires the design of our new birthmark for Android apps, namely, feature view graph, which captures users' navigation behavior across app views. Our experimental results demonstrate that this birthmark can characterize Android apps from a higher level abstraction, making it resilient to code obfuscation. ViewDroid can detect repackaged apps at a large scale, both effectively and efficiently. Our experiments also show that the false positive and false negative rates of ViewDroid are both very low.

/pdf/viewdroid-towards-obfuscation-resilient-mobile-application-3thmhkuhdp.pdf

ViewDroid: towards obfuscation-resilient mobile application repackaging detection

Because it is not hard to reverse engineer the Dalvik bytecode used in the Dalvik virtual machine, Android application repackaging has become a serious problem. With repackaging, a plagiarist can simply steal others’ code violating the intellectual property of the developers. More seriously, after repackaging, popular apps can become the carriers of malware, adware or spy-ware for wide spreading. To maintain a healthy app market, several detection algorithms have been proposed recently, which can catch some types of repackaged apps in various markets efficiently. However, they are generally lack of valid analysis on their effectiveness. After analyzing these approaches, we find simple obfuscation techniques can potentially cause false negatives, because they change the main characteristics or features of the apps that are used for similarity detections. In practice, more sophisticated obfuscation techniques can be adopted (or have already been performed) in the context of mobile apps. We envision this obfuscation based repackaging will become a phenomenon due to the arms race between repackaging and its detection. To this end, we propose a framework to evaluate the obfuscation resilience of repackaging detection algorithms comprehensively. Our evaluation framework is able to perform a set of obfuscation algorithms in various forms on the Dalvik bytecode. Our results provide insights to help gauge both broadness and depth of algorithms’ obfuscation resilience. We applied our framework to conduct a comprehensive case study on AndroGuard, an Android repackaging detector proposed in Black-hat 2011. Our experimental results have demonstrated the effectiveness and stability of our framework.

/pdf/a-framework-for-evaluating-mobile-app-repackaging-detection-51wo62wwog.pdf

Heqing Huang

Papers

Protecting Intellectual Property of Deep Neural Networks with Watermarking

Commandersong: a systematic approach for practical adversarial voice recognition

Finding unknown malice in 10 seconds: mass vetting for new threats at the Google-play scale

ViewDroid: towards obfuscation-resilient mobile application repackaging detection

A Framework for Evaluating Mobile App Repackaging Detection Algorithms