Modern OS kernels are bloated to support diverse hardware and features, yet most deployments use only a small subset. This paper advances configuration-based kernel debloating to reduce attack surface and code size in systems with known workloads. We present Tracie,a trace-based specialization tool that improves precision through better workload attribution and trace-to-config mapping, achieving a 21.67% reduction—outperforming prior work by 8%—and removing 1 additional CVEs. We also introduce Dice, a novel trace-free, dependency-aware debloating approach that iteratively prunes the configuration graph, achieving 35.59% reduction and eliminating 10 more CVEs. Finally, we empirically analyze the Linux kconfig system and show that structural limitations—such as coarse granularity and rigid dependencies—constrain the effectiveness of configuration-based debloating.

In this paper, we systematically evaluate the robustness of multi-exit language models against adversarial slowdown. To audit their robustness, we design a slowdown attack that generates natural adversarial text bypassing early-exit points. We use the resulting WAFFLE attack as a vehicle to conduct a comprehensive evaluation of three multi-exit mechanisms with the GLUE benchmark against adversarial slowdown. We then show our attack significantly reduces the computational savings provided by the three methods in both white-box and black-box settings. The more complex a mechanism is, the more vulnerable it is to adversarial slowdown. We also perform a linguistic analysis of the perturbed text inputs, identifying common perturbation patterns that our attack generates, and comparing them with standard adversarial text attacks. Moreover, we show that adversarial training is ineffective in defeating our slowdown attack, but input sanitization with a conversational model, e.g., ChatGPT, can remove perturbations effectively. This result suggests that future work is needed for developing efficient yet robust multi-exit models. Our code is available at: https://github.com/ztcoalson/WAFFLE

Super-Resolution Convolutional Neural Networks (SRCNNs) with their ability to generate highresolution images from low-resolution counterparts, exacerbate the privacy concerns emerging from automated Convolutional Neural Networks (CNNs)-based image classifiers. In this work, we hypothesize and empirically show that adversarial examples learned over CNN image classifiers can survive processing by SRCNNs and lead them to generate poor quality images that are hard to classify correctly. We demonstrate that a user with a small CNN is able to learn adversarial noise without requiring any customization for SRCNNs and thwart the privacy threat posed by a pipeline of SRCNN and CNN classifiers (95.8% fooling rate for Fast Gradient Sign with ε = 0.03). We evaluate the survivability of adversarial images generated in both black-box and white-box settings and show that black-box adversarial learning (when both CNN classifier and SRCNN are unknown) is at least as effective as white-box adversarial learning (when only CNN classifier is known). We also assess our hypothesis on adversarial robust CNNs and observe that the supper-resolved white-box adversarial examples can fool these CNNs more than 71.5% of the time.

Abstract End users are increasingly using trigger-action platforms like If-This-Then-That (IFTTT) to create applets to connect smart-home devices and services. However, there are inherent implicit risks in using such applets—even non-malicious ones—as sensitive information may leak through their use in certain contexts (e.g., where the device is located, who can observe the resultant action). This work aims to understand to what extent end users can assess this implicit risk. More importantly we explore whether usage context makes a difference in end-users’ perception of such risks. Our work complements prior work that has identified the impact of usage context on expert evaluation of risks in IFTTT by focusing the impact of usage context on end-users’ risk perception. Through a Mechanical Turk survey of 386 participants on 49 smart-home IFTTT applets, we found that participants have a nuanced view of contextual factors and that different values for contextual factors impact end-users’ risk perception differently. Further, our findings show that nudging the participants to think about different usage contexts led them to think deeper about the associated risks and raise their concern scores.

Modern embedded real-time systems (RTS) are increasingly facing more security threats than the past. A simplistic straightforward integration of security mechanisms might not be able to guarantee thesafetyand predictability of such systems. In this article, we focus on integrating security mechanisms into RTS (especiallylegacyRTS). We introduceContego-C, an analytical model to integrate security tasks into RTS that will allow system designers to improve the security posture without affecting temporal and control constraints of the existing real-time control tasks. We also define ametric(named tightness of periodic monitoring) to measure the effectiveness of such integration. We demonstrate our ideas using a proof-of-concept implementation on an ARM-based rover platform and show that Contego-C can improve security without degrading control performance.

Safety-critical networks often have stringent real-time requirements; they must also be resilient to failures. In this paper, we propose the RealFlow framework that uses commodity software-defined networks (SDNs) to realize networks with end-to-end timing guarantees, while also: (a) increasing resiliency against link/switch failures and (b) increasing network utilization. The use of SDNs in this space also improves the management capabilities of the system due to the global visibility into the network. RealFlow is implemented as a northbound SDN controller application compatible with standard OpenFlow protocols with little to no runtime overheads. We demonstrate feasibility on a real hardware testbed (Pica8 SDN switches+Raspberry Pi endhosts) and a practical avionics case study. Our evaluations show that RealFlow can accommodate 63% more network flows with safety-critical guarantees when compared to current designs and up to 18% when link resiliency (via backup paths) is also considered.

Abstract Image hosting platforms are a popular way to store and share images with family members and friends. However, such platforms typically have full access to images raising privacy concerns. These concerns are further exacerbated with the advent of Convolutional Neural Networks (CNNs) that can be trained on available images to automatically detect and recognize faces with high accuracy. Recently, adversarial perturbations have been proposed as a potential defense against automated recognition and classification of images by CNNs. In this paper, we explore the practicality of adversarial perturbation-based approaches as a privacy defense against automated face recognition. Specifically, we first identify practical requirements for such approaches and then propose two practical adversarial perturbation approaches – (i) learned universal ensemble perturbations (UEP), and (ii) k-randomized transparent image overlays (k-RTIO) that are semantic adversarial perturbations. We demonstrate how users can generate effective transferable perturbations under realistic assumptions with less effort. We evaluate the proposed methods against state-of-theart online and offline face recognition models, Clarifai.com and DeepFace, respectively. Our findings show that UEP and k-RTIO respectively achieve more than 85% and 90% success against face recognition models. Additionally, we explore potential countermeasures that classifiers can use to thwart the proposed defenses. Particularly, we demonstrate one effective countermeasure against UEP.

We propose HYDRA-C, a design-time evaluation framework for integrating monitoring mechanisms in multicore real-time systems (RTS). Our goal is to ensure that security (or other monitoring) mechanisms execute in a "continuous" manner - i.e., as often as possible, across cores. This is to ensure that any such mechanisms run with few interruptions, if any. HYDRA-C is intended to allow designers of RTS to integrate monitoring mechanisms without perturbing existing timing properties or execution orders. We demonstrate the framework using a proofof-concept implementation with intrusion detection mechanisms as security tasks. We develop and use both, (a) a custom intrusion detection system (IDS) as well as (b) Tripwire - an open source data integrity checking tool. We compare the performance of HYDRA-C with a state-of-the-art multicore RT security integration approach and find that our method does not impact the schedulability and, on average, can detect intrusions 19.05% faster without impacting the performance of RT tasks.

Vanilla CNNs, as uncalibrated classifiers, suffer from classifying out-of-distribution (OOD) samples nearly as confidently as in-distribution samples. To tackle this challenge, some recent works have demonstrated the gains of leveraging available OOD sets for training end-to-end calibrated CNNs. However, a critical question remains unanswered in these works: how to differentiate OOD sets for selecting the most effective one(s) that induce training such CNNs with high detection rates on unseen OOD sets? To address this pivotal question, we provide a criterion based on generalization errors of Augmented-CNN, a vanilla CNN with an added extra class employed for rejection, on in-distribution and unseen OOD sets. However, selecting the most effective OOD set by directly optimizing this criterion incurs a huge computational cost. Instead, we propose three novel computationally-efficient metrics for differentiating between OOD sets according to their level of in-distribution sub-manifolds. We empirically verify that the most protective OOD sets -- selected according to our metrics -- lead to A-CNNs with significantly lower generalization errors than the A-CNNs trained on the least protective ones. We also empirically show the effectiveness of a protective OOD set for training well-generalized confidence-calibrated vanilla CNNs. These results confirm that 1) all OOD sets are not equally effective for training well-performing end-to-end models (i.e., A-CNNs and calibrated CNNs) for OOD detection tasks and 2) the protection level of OOD sets is a viable factor for recognizing the most effective one. Finally, across the image classification tasks, we exhibit A-CNN trained on the most protective OOD set can also detect black-box FGS adversarial examples as their distance (measured by our metrics) is becoming larger from the protected sub-manifolds.

We demonstrate the presence of a novel scheduler side-channel in preemptive, fixed-priority real-time systems (RTS); examples of such systems can be found in automotive systems, avionic systems, power plants and industrial control systems among others. This side-channel can leak important timing information such as the future arrival times of real-time tasks. This information can then be used to launch devastating attacks, two of which are demonstrated here (on real hardware platforms). Note that it is not easy to capture this timing information due to runtime variations in the schedules, the presence of multiple other tasks in the system and the typical constraints (e.g., deadlines) in the design of RTS. Our ScheduLeak algorithms demonstrate how to effectively exploit this side-channel. A complete implementation is presented on real operating systems (in Real-time Linux and FreeRTOS). Timing information leaked by ScheduLeak can significantly aid other, more advanced, attacks in better accomplishing their goals.

The increased capabilities of modern real-time systems (RTS) expose them to various security threats. Recently, frameworks that integrate security tasks without perturbing the real-time tasks have been proposed, but they only target single core systems. However, modern RTS are migrating towards multicore platforms. This makes the problem of integrating security mechanisms more complex, as designers now have multiple choices for where to allocate the security tasks. In this paper we propose HYDRA, a design space exploration algorithm that finds an allocation of security tasks for multicore RTS using the concept of opportunistic execution. HYDRA allows security tasks to operate with existing real-time tasks without perturbing system parameters or normal execution patterns, while still meeting the desired monitoring frequency for intrusion detection. Our evaluation uses a representative real-time control system (along with synthetic task sets for a broader exploration) to illustrate the efficacy of HYDRA.

Despite the isolation mechanisms that are available to cloud service providers, like virtual machines and containers, the problem of side-channel vulnerabilities due to shared caches and multicore processors remains a threat. We present a hardware-software mechanism that improves the isolation of cloud processes in the presence of shared caches on multicore chips. Our technique can enable cache-side-channel free computing for Linux-based containers and virtual machines by com-bining the Intel CAT architecture that enables cache partitioning with novel scheduling techniques and state cleansing mechanisms. We evaluate our system using a CPU-bound workload and demonstrate cache-side-channel-free computation that is correct by construction. Our system allows Simultaneous Multithreading to remain enabled and does not require application level changes.

Cyber-Physical Systems (CPS) are becoming increasingly critical for the well-being of society (e.g., electricity generation and distribution, water treatment, implantable medical devices etc.). While the convergence of computing, communications and physical control in such systems provides benefits in terms of efficiency and convenience, the attack surface resulting from this convergence poses unique security and privacy challenges. These systems represent the new frontier for cyber risk. CPS-SPC is an annual forum, in its 3rd edition this year, that aims to provide a focal point for the research community to begin addressing the security and privacy challenges of CPS in a comprehensive and multidisciplinary manner and, in tandem with other efforts, build a comprehensive research road map.

A common security practice used to deal with a password breach is locking user accounts and sending out an email to tell users that they need to reset their password to unlock their account. This paper evaluates the effectiveness of this security practice based on the password reset email that LinkedIn sent out around May 2016, and through an online survey conducted on 249 LinkedIn users who received that email. Our evaluation shows that only about 46% of the participants reset their passwords. The mean time taken to reset password was 26.3 days, revealing that a significant proportion of the participants reset their password a few weeks, or even months after first receiving the email. Our findings suggest that more effective persuasive measures need to be added to convince users to reset their password in a timely manner, and further reduce the risks associated with delaying password resets.

Embedded real-time systems (RTS) are pervasive. Many modern RTS are exposed to unknown security flaws, and threats to RTS are growing in both number and sophistication. However, until recently, cyber-security considerations were an afterthought in the design of such systems. Any security mechanisms integrated into RTS must (a) co-exist with the real-time tasks in the system and (b) operate without impacting the timing and safety constraints of the control logic. We introduce Contego, an approach to integrating security tasks into RTS without affecting temporal requirements. Contego is specifically designed for legacy systems, viz., the real-time control systems in which major alterations of the system parameters for constituent tasks is not always feasible. Contego combines the concept of opportunistic execution with hierarchical scheduling to maintain compatibility with legacy systems while still providing flexibility by allowing security tasks to operate in different modes. We also define a metric to measure the effectiveness of such integration. We evaluate Contego using synthetic workloads as well as with an implementation on a realistic embedded platform (an open-source ARM CPU running real-time Linux).

Numerous event-based probing methods exist for cloud computing environments allowing a hypervisor to gain insight into guest activities. Such event-based probing has been shown to be useful for detecting attacks, system hangs through watchdogs, and for inserting exploit detectors before a system can be patched, among others. Here, we illustrate how to use such probing for trustworthy logging and highlight some of the challenges that existing event-based probing mechanisms do not address. Challenges include ensuring a probe inserted at given address is trustworthy despite the lack of attestation available for probes that have been inserted dynamically. We show how probes can be inserted to ensure proper logging of every invocation of a probed instruction. When combined with attested boot of the hypervisor and guest machines, we can ensure the output stream of monitored events is trustworthy. Using these techniques we build a trustworthy log of certain guest-system-call events. The log powers a cloud-tuned Intrusion Detection System (IDS). New event types are identified that must be added to existing probing systems to ensure attempts to circumvent probes within the guest appear in the log. We highlight the overhead penalties paid by guests to increase guarantees of log completeness when faced with attacks on the guest kernel. Promising results (less that 10% for guests) are shown when a guest relaxes the trade-off between log completeness and overhead. Our demonstrative IDS detects common attack scenarios with simple policies built using our guest behavior recording system.

Real-time systems (RTS) require end-to-end delay guarantees for the delivery of network packets. In this paper, we propose a framework to reduce the management and integration overheads for such real-time (RT) network flows by leveraging the capabilities of software-defined networking (SDN) - capabilities that include global visibility and management of the network. Given the specifications of flows that must meet hard real-time requirements, our framework synthesizes paths through the network. To guarantee that these flows meet both, their bandwidth and end-to-end timing requirements, our framework solves a multi-constraint optimization problem using a heuristic algorithm. We use exhaustive emulations and experiments on hardware switches to demonstrate our techniques and feasibility of our approach. As a result of this work, SDNs become “delay-aware” and thus can be adapted for use in safety-critical and other delay-sensitive applications.

Home automation has become increasingly popular, with new interconnected products being introduced on a regular basis. While the benefits of these devices are tantalizing, end users may not fully understand the complexities of setting up these devices, become frustrated with the process, or have incorrect installations. We performed an exploratory study to understand the barriers that they face in actually setting up these devices. Participants faced multiple barriers, some of which were insurmountable. Our work indicates that current home automation devices run contrary to the perception that smart homes devices are “plug-and-play”.

The Second International Workshop on Cyber-Physical Systems Security and PrivaCy (CPS-SPC'16) is being held in conjunction with the 23rd ACM CCS Conference. This second edition follows a successful workshop held with ACM CCS in 2015. The workshop was motivated by several observations. First, cyber-physical systems represent the new frontier for cyber risk. The attack surface imposed by the convergence of computing, communications and physical control represents unique challenges for security researchers and practitioners. Second, majority of the published literature addressing the security and privacy of CPS reflect a field still in its infancy. As such, the overall principles, models, and theories for securing CPS have not yet emerged. Third, the organizers of this workshop strongly felt that a premiere forum associated with a premiere conference was needed for rapidly publishing diverse, multidisciplinary in-progress work on the security and privacy of CPS and galvanizing the research community. The set of accepted papers reflect this vision. We have organized an exciting program for this workshop and look forward to active participation in this and future workshops.

Due to physical isolation as well as use of proprietary hardware and protocols, traditional real-time systems (RTS) were considered to be invulnerable to security breaches and external attacks. This assumption is being challenged by recent attacks that highlight vulnerabilities in RTS. Besides, a straightforward integration of security mechanisms might compromise the safety and predictability guarantees of such systems. In this paper, we focus on integrating security mechanisms into RTS (especially legacy RTS) and define a metric to measure the effectiveness of such integration. We combine opportunistic execution with hierarchical scheduling to maintain compatibility with legacy systems while still providing flexibility. The proposed approach is shown to increase the security posture of RTS without impacting their temporal (and hence, safety) constraints.

The First International Workshop on Cyber-Physical Systems Security and PrivaCy (CPS-SPC) is being held in conjunction with the 22nd ACM CCS Conference. The workshop was motivated by several observations. First, cyber-physical systems represent the new frontier for cyber risk. The attack surface imposed by the convergence of computing, communications and physical control represents unique challenges for security researchers and practitioners. Second, majority of the published literature addressing the security and privacy of CPS reflect a field still in its infancy. As such, the overall principles, models, and theories for securing CPS have not yet emerged. Third, the organizers of this workshop strongly felt that a premiere forum associated with a premiere conference was needed for rapidly publishing diverse, multidisciplinary in-progress work on the security and privacy of CPS and galvanizing the research community. The set of accepted papers reflect this vision. Papers span cyber and control-theoretic foundations, intrusion detection, forensics management, vulnerability analysis and elimination, and field studies. We have organized an exciting program for this workshop and look forward to active participation in this and future workshops.

Traditionally real-time systems and security have been considered as separate domains. Recent attacks on various systems with real-time properties have shown the need for a redesign of such systems to include security as a first class principle. In this paper, we propose a general model for capturing security constraints between tasks in a real-time system. This model is then used in conjunction with real-time scheduling algorithms to prevent the leakage of information via storage channels on implicitly shared resources. We expand upon a mechanism to enforce these constraints viz., cleaning up of shared resource state, and provide schedulability conditions based on fixed priority scheduling with both preemptive and non-preemptive tasks. We perform extensive evaluations, both theoretical and experimental, the latter on a hardware-in-the-loop simulator of an unmanned aerial vehicle (UAV) that executes on a demonstration platform.

Real-time systems (RTS) were typically considered to be invulnerable to external attacks, mainly due to their use of proprietary hardware and protocols, as well as physical isolation. As a result, RTS and security have traditionally been separate domains. These assumptions are being challenged by a series of recent events that highlight the vulnerabilities in RTS. In this paper we focus on integrating security as a first class principle in the design of RTS: we show that certain security requirements can be specified as real-time scheduling constraints. Using information leakage as a motivating problem, we illustrate our techniques with fixed-priority (FP) real-time schedulers. We evaluate our approach and discuss tradeoffs. Our evaluation shows that many real-time task sets can be scheduled under the proposed constraints without significant performance impact.

Security-as-a-Service (SecaaS) is gaining popularity, with cloud-based anti-spam and anti-virus leading the way. In this work we look at key management as a security service and focus on group key management witha central group key manager. Specifically, we analyze are writing logic model of a ZooKeeper-based group key management service specified in Maude and study its tolerance to faults and performance as it scales to service larger groups using the PVeStA statistical model checking tool.

Virtual appliances (VAs) are ready-to-use virtual machine images that are configured for specific purposes. For example, a virtual machine image that contains all the software necessary to develop and host a JSP-based website is typically available as a "Java Web Starter" VA. Currently there are many VA repositories from which users can download VAs and instantiate them on Infrastructure-as-a-Service (IaaS) clouds, allowing them to quickly launch their services. This marketplace, however, lacks adequate mechanisms that allow users to a priori assess whether a specific VA is really configured with the software that it is expected to be configured with. This paper evaluates the integrity of software packages installed on real-world VAs, through the use of a software whitelist-based framework, and finds that indeed there is a lot of variance in the software integrity of packages across VAs. Analysis of 151 Amazon VAs using this framework shows that about 9% of real-world VAs have significant numbers of software packages that contain unknown files, making them potentially untrusted. Virus scanners flagged just half of the VAs in that 9% as malicious, demonstrating that virus scanning alone is not sufficient to help users select a trustable VA and that a priori software integrity assessment has a role to play.

In today's IaaS clouds users only get a logical view of the underlying network and have limited control. Delegating more control to end users would be beneficial but would also raise security concerns for the provider. Emerging Software Defined Networking (SDN) technologies have the capabilities to facilitate delegation of network controls and provide some level of network abstractions to end users. However, any delegation solution should try to balance the level of controls delegated to end users with the security constraints of the provider. In this paper, we propose a SDN-based framework to facilitate delegation of some network controls to end users, providing the means to monitor and configure their own slices of the underlying networks. Using two instantiations of this framework, we illustrate the tradeoffs between security and the level of network abstractions provided to end users.

Integrated role-based access control (RBAC) and attribute-based access control (ABAC) is emerging as a promising paradigm. This paper proposes a framework that uses attribute-based policies to create a more traditional RBAC model. RBAC has been widely used, but has weaknesses: it is labor-intensive and time-consuming to build a model instance, and a pure RBAC system lacks flexibility to efficiently adapt to changing users, objects, and security policies. Particularly, it is impractical to manually make (and maintain) user to role assignments and role to permission assignments in industrial context characterized by a large number of users and/or security objects. ABAC has features complimentary to RBAC, and merging RBAC and ABAC has become an important research topic. This paper proposes a new approach to integrating ABAC with RBAC, by modeling RBAC in two levels. The aboveground level is a standard RBAC model extended with "environment". This level retains the simplicity of RBAC, supporting RBAC model verification/review. The "underground" level is used to represent security knowledge in terms of attribute-based policies, which automatically create the simple RBAC model in the aboveground level. These attribute-based policies bring to RBAC the advantages of ABAC: they are easy to build and easy to adapt to changes. Using this framework, we tackle the problem of permission assignment for large scale applications. This model is motivated by the characteristics and requirements of industrial control systems, and reflects in part certain approaches and practices common in the industry.

Attribute-Based Messaging (ABM) enables messages to be addressed using attributes of recipients rather than an explicit list of recipients. Such messaging offers benefits of efficiency, exclusiveness, and intensionality, but faces challenges in access control and confidentiality. In this article we explore an approach to intraenterprise ABM based on providing access control and confidentiality using information from the same attribute database exploited by the addressing scheme. We show how to address three key challenges. First, we demonstrate a manageable access control system based on attributes. Second, we demonstrate use of attribute-based encryption to provide end-to-end confidentiality. Third, we show that such a system can be efficient enough to support ABM for mid-size enterprises. Our implementation can dispatch confidential ABM messages approved by XACML policy review for an enterprise of at least 60,000 users with only seconds of latency.

In distributed systems users need the ability to share sensitive content with multiple other recipients based on their ability to satisfy arbitrary policies. One such system is electricity grids where finegrained sensor data sharing holds the potential for increased reliability and efficiency. However, effective data sharing requires technical solutions that support flexible access policies, for example, sharing more data when the grid is unstable. In such systems, both the messages and policies are sensitive and, therefore, they need to kept be secret. Furthermore, to allow for such a system to be secure and usable in the presence of untrusted object stores and relays it must be resilient in the presence of active adversaries and provide efficient key management. While several of these properties have been studied in the past we address a new problem in the area of policy based encryption in that we develop a solution with all of these capabilities. We develop a Policy and Key Encapsulation Mechanism -- Data Encapsulation Mechanism (PKEM-DEM) encryption scheme that is a generic construction secure against adaptive chosen ciphertext attacks and develop a Policy Based Encryption System (PBES) using this scheme that provides these capabilities. We provide an implementation of PBES and measure its performance.

Attribute based messaging (ABM) enables message senders to dynamically create a list of recipients based on their attributes as inferred from an enterprise database. Such targeted messaging can reduce unnecessary communications and enhance privacy, but faces challenges in access control. In this paper, we explore an approach to ABM based on deriving access control information from the same attribute database exploited by the addressing scheme. We show how to address three key challenges. First, we demonstrate a manageable access control system based on attributes. Second we show how this can be used with existing messaging systems to provide a practical deployment strategy. Third, we show that such a system can be efficient enough to support ABM for mid-size enterprises. Our implementation can dispatch ABM messages approved by XACML review for an enterprise of at least 60,000 users with only seconds of latency

To date, most solutions proposed for secure routing in mobile ad-hoc networks (MANETs), assume that secure associations between pairs of nodes can be established on-line; e.g., by a trusted third party, by distributed trust establishment. However, establishing such security associations, with or without trusted third parties, requires reliance on routing layer security. In this paper, we eliminate this apparent cyclic dependency between security services and secure routing in MANETs and show how to bootstrap security for the routing layer. We use the notion of statistically unique and cryptographically verifiable (SUCV) identifiers to implement a secure binding between IP addresses and keys that is independent of any trusted security service. We illustrate our solution with the dynamic source routing (DSR) protocol and compare it with other solutions for secure routing.