arXiv subjects

Hidayet Aksu

Publications and source records attributed to Hidayet Aksu.

At least 19 recordsLinked to original sources

arXiv2026

Measuring Obedience to Authority Across Large Language Models with the Milgram Paradigm

Large language models (LLMs) are increasingly deployed as agents that operate equipment, execute instructions, and act inside institutional hierarchies, raising a question social psychology answered for humans six decades ago: how far will an agent escalate a harmful action when a legitimate authority insists? We port Milgram's obedience paradigm to LLMs as a standardized, fully scripted, replicable probe: the model plays the Teacher, a deterministic harness plays Experimenter and Learner from paraphrased versions of Milgram's scripts (30 shock levels, 15-450 V; graded protests; the four standardized prods), and the outcome of a session is the breakoff voltage. We measure obedience profiles, empirical breakoff distributions over a battery of six conditions, for 42 models from 19 families (4848 sessions, 102511 logged decision turns). We find that (i) obedience is extremely heterogeneous, with baseline full-obedience rates spanning 0%-100% (census mean 42.9%; human anchor 65%). (ii) Profiles are model-specific and stable: split-half verification separates same-model from cross-model comparisons at AUC = 0.885. (iii) Situational sensitivity is selective: scripted peer defiance shifts obedience in the human direction, learner proximity trends the same way without reaching significance, and removing the authority's physical presence, one of the strongest human levers, trends in the opposite direction, also without reaching significance. (iv) Declaring the scenario fictional raises obedience, whereas moving the decision from a typed action line to a native tool call, or granting a modest thinking budget, lowers it sharply. (v) Unlike single-token fingerprints, obedience profiles do not recover model lineage: obedience identifies the checkpoint but not its ancestry, consistent with safety post-training overwriting lineage priors.

cs.CR↗

arXiv2026

Maestro Order: A Model-Agnostic Orchestration Harness

A single forward pass of a capable model is a fast, fluent, and unreliable problem-solver: it is right often enough to be useful and wrong often enough to be dangerous; in language models, such confident errors are known as hallucinations. We present Maestro Order, a model-agnostic orchestration harness that turns unreliable solvers into reliable problem-solving systems by composing them according to four structural primitives (decompose, ensemble, verify, and recurse) and a budget-aware controller that decides where to spend compute. The harness treats any model as a black-box base solver behind a uniform interface, layers a verifier ensemble whose discrimination is measured online, and allocates verification and voting to the stages with the highest marginal reliability per unit cost. We give the architecture, the message and state schema, the controller algorithm, and the engineering that makes it deterministic, observable, and fault-tolerant. We then specify an evaluation methodology (reliability at fixed cost, coverage, calibration, and ablations) and report results from a faithful Monte Carlo simulation of the harness over a parameterized solver/verifier model. The simulation reproduces the predicted laws quantitatively: verification amplifies reliability geometrically (e.g. $0.55\to0.98$ with two gates, $\to0.999$ with four), voting helps only above chance and is limited by shared errors, and a budget-aware controller reaches a target reliability at a small fraction of the cost of voting alone by selecting the cheapest mechanism for each regime. We close with failure modes (verifier gaming, correlated errors, and decomposition error compounding) and concrete guidance: build robust checkers, diversify solvers, and let the controller put compute where the information is.

cs.CR↗

arXiv2026

Odds Law: The Decomposition Algebra On How Intelligence Organizes Itself to Solve Difficult Problems Reliably

We ask a structural question: given unreliable elementary problem-solvers, what organizations of them solve hard problems reliably, and what are the limits? We develop a $decomposition~algebra$: elementary solvers are morphisms in a stochastic category, and four combinators (sequential composition, parallel ensembling, verification gating, and recursive reduction) generate the space of compound solvers. We equip this algebra with two homomorphisms, a $reliability$ valuation into the ordered monoid $([0,1],\le)$ and a $cost$ valuation into a commutative semiring, and we derive the composition laws that govern how reliability flows through structure. Our central results are (i) a $verification~odds~law$ (the result that names this report), showing that a verification gate multiplies the odds of correctness by the verifier's likelihood ratio $Λ$, so that $k$ conditionally independent gates yield geometric amplification; (ii) a $reliability~amplification~theorem$, giving target reliability $1-δ$ at $O(\log 1/δ)$ verification depth whenever $Λ>1$; and (iii) a $threshold~dichotomy$: above the critical parameters reliability can be driven arbitrarily close to one at logarithmic cost, while at or below them no amplification is possible. We then show that $self-organization$ is the least fixed point of a monotone improvement operator on the complete lattice of strategies, and that this fixed point equalizes marginal log-odds gain per unit cost. Finally, we prove matching limits: an information ceiling bounds per-gate amplification by a divergence quantity; shared error causes create a strictly positive voting floor, so diversity is $necessary$ for unbounded amplification. Reliability, in short, is neither free nor magical: it is bought with independent information, arranged by composition, and bounded by the verifier.

cs.CR↗

arXiv2025

Identification of Wearable Devices with Bluetooth

With wearable devices such as smartwatches on the rise in the consumer electronics market, securing these wearables is vital. However, the current security mechanisms only focus on validating the user not the device itself. Indeed, wearables can be (1) unauthorized wearable devices with correct credentials accessing valuable systems and networks, (2) passive insiders or outsider wearable devices, or (3) information-leaking wearables devices. Fingerprinting via machine learning can provide necessary cyber threat intelligence to address all these cyber attacks. In this work, we introduce a wearable fingerprinting technique focusing on Bluetooth classic protocol, which is a common protocol used by the wearables and other IoT devices. Specifically, we propose a non-intrusive wearable device identification framework which utilizes 20 different Machine Learning (ML) algorithms in the training phase of the classification process and selects the best performing algorithm for the testing phase. Furthermore, we evaluate the performance of proposed wearable fingerprinting technique on real wearable devices, including various off-the-shelf smartwatches. Our evaluation demonstrates the feasibility of the proposed technique to provide reliable cyber threat intelligence. Specifically, our detailed accuracy results show on average 98.5%, 98.3% precision and recall for identifying wearables using the Bluetooth classic protocol.

cs.CR↗

arXiv2023

Summary Reports Optimization in the Privacy Sandbox Attribution Reporting API

The Privacy Sandbox Attribution Reporting API has been recently deployed by Google Chrome to support the basic advertising functionality of attribution reporting (aka conversion measurement) after deprecation of third-party cookies. The API implements a collection of privacy-enhancing guardrails including contribution bounding and noise injection. It also offers flexibility for the analyst to allocate the contribution budget. In this work, we present methods for optimizing the allocation of the contribution budget for summary reports from the Attribution Reporting API. We evaluate them on real-world datasets as well as on a synthetic data model that we find to accurately capture real-world conversion data. Our results demonstrate that optimizing the parameters that can be set by the analyst can significantly improve the utility achieved by querying the API while satisfying the same privacy bounds.

cs.CR↗

arXiv2020

KRATOS: Multi-User Multi-Device-Aware Access Control System for the Smart Home

In a smart home system, multiple users have access to multiple devices, typically through a dedicated app installed on a mobile device. Traditional access control mechanisms consider one unique trusted user that controls the access to the devices. However, multi-user multi-device smart home settings pose fundamentally different challenges to traditional single-user systems. For instance, in a multi-user environment, users have conflicting, complex, and dynamically changing demands on multiple devices, which cannot be handled by traditional access control techniques. To address these challenges, in this paper, we introduce Kratos, a novel multiuser and multi-device-aware access control mechanism that allows smart home users to flexibly specify their access control demands. Kratos has three main components: user interaction module, backend server, and policy manager. Users can specify their desired access control settings using the interaction module which are translated into access control policies in the backend server. The policy manager analyzes these policies and initiates negotiation between users to resolve conflicting demands and generates final policies. We implemented Kratos and evaluated its performance on real smart home deployments featuring multi-user scenarios with a rich set of configurations (309 different policies including 213 demand conflicts and 24 restriction policies). These configurations included five different threats associated with access control mechanisms. Our extensive evaluations show that Kratos is very effective in resolving conflicting access control demands with minimal overhead and robust against different attacks.

cs.CR↗

arXiv2020

Peek-a-Boo: I see your smart home activities, even encrypted!

A myriad of IoT devices such as bulbs, switches, speakers in a smart home environment allow users to easily control the physical world around them and facilitate their living styles through the sensors already embedded in these devices. Sensor data contains a lot of sensitive information about the user and devices. However, an attacker inside or near a smart home environment can potentially exploit the innate wireless medium used by these devices to exfiltrate sensitive information from the encrypted payload (i.e., sensor data) about the users and their activities, invading user privacy. With this in mind,in this work, we introduce a novel multi-stage privacy attack against user privacy in a smart environment. It is realized utilizing state-of-the-art machine-learning approaches for detecting and identifying the types of IoT devices, their states, and ongoing user activities in a cascading style by only passively sniffing the network traffic from smart home devices and sensors. The attack effectively works on both encrypted and unencrypted communications. We evaluate the efficiency of the attack with real measurements from an extensive set of popular off-the-shelf smart home IoT devices utilizing a set of diverse network protocols like WiFi, ZigBee, and BLE. Our results show that an adversary passively sniffing the traffic can achieve very high accuracy (above 90%) in identifying the state and actions of targeted smart home devices and their users. To protect against this privacy leakage, we also propose a countermeasure based on generating spoofed traffic to hide the device states and demonstrate that it provides better protection than existing solutions.

cs.CR↗

arXiv2019

A System-level Behavioral Detection Framework for Compromised CPS Devices: Smart-Grid Case

Cyber-Physical Systems (CPS) play a significant role in our critical infrastructure networks from power-distribution to utility networks. The emerging smart-grid concept is a compelling critical CPS infrastructure that relies on two-way communications between smart devices to increase efficiency, enhance reliability, and reduce costs. However, compromised devices in the smart grid poses several security challenges. Consequences of propagating fake data or stealing sensitive smart grid information via compromised devices are costly. Hence, early behavioral detection of compromised devices is critical for protecting the smart grid's components and data. To address these concerns, in this paper, we introduce a novel and configurable system-level framework to identify compromised smart grid devices. The framework combines system and function call tracing techniques with signal processing and statistical analysis to detect compromised devices based on their behavioral characteristics. We measure the efficacy of our framework with a realistic smart grid substation testbed that includes both resource-limited and resource-rich devices. In total, using our framework, we analyze six different types of compromised device scenarios with different resources and attack payloads. To the best of our knowledge, the proposed framework is the first in detecting compromised CPS smart grid devices with system and function-level call tracing techniques. The experimental results reveal an excellent rate for the detection of compromised devices. Specifically, performance metrics include accuracy values between 95% and 99% for the different attack scenarios. Finally, the performance analysis demonstrates that the use of the proposed framework has minimal overhead on the smart grid devices' computing resources.

cs.CR↗

arXiv2019

A Context-aware Framework for Detecting Sensor-based Threats on Smart Devices

Sensors (e.g., light, gyroscope, accelerometer) and sensing-enabled applications on a smart device make the applications more user-friendly and efficient. However, the current permission-based sensor management systems of smart devices only focus on certain sensors and any App can get access to other sensors by just accessing the generic sensor Application Programming Interface (API). In this way, attackers can exploit these sensors in numerous ways: they can extract or leak users' sensitive information, transfer malware, or record or steal sensitive information from other nearby devices. In this paper, we propose 6thSense, a context-aware intrusion detection system which enhances the security of smart devices by observing changes in sensor data for different tasks of users and creating a contextual model to distinguish benign and malicious behavior of sensors. 6thSense utilizes three different Machine Learning-based detection mechanisms (i.e., Markov Chain, Naive Bayes, and LMT). We implemented 6thSense on several sensor-rich Android-based smart devices (i.e., smart watch and smartphone) and collected data from typical daily activities of 100 real users. Furthermore, we evaluated the performance of 6thSense against three sensor-based threats: (1) a malicious App that can be triggered via a sensor, (2) a malicious App that can leak information via a sensor, and (3) a malicious App that can steal data using sensors. Our extensive evaluations show that the 6thSense framework is an effective and practical approach to defeat growing sensor-based threats with an accuracy above 96% without compromising the normal functionality of the device. Moreover, our framework reveals minimal overhead.

cs.CR↗

arXiv2019

Aegis: A Context-aware Security Framework for Smart Home Systems

Our everyday lives are expanding fast with the introduction of new Smart Home Systems (SHSs). Today, a myriad of SHS devices and applications are widely available to users and have already started to re-define our modern lives. Smart home users utilize the apps to control and automate such devices. Users can develop their own apps or easily download and install them from vendor-specific app markets. App-based SHSs offer many tangible benefits to our lives, but also unfold diverse security risks. Several attacks have already been reported for SHSs. However, current security solutions consider smart home devices and apps individually to detect malicious actions rather than the context of the SHS as a whole. The existing mechanisms cannot capture user activities and sensor-device-user interactions in a holistic fashion. To address these issues, in this paper, we introduce Aegis, a novel context-aware security framework to detect malicious behavior in a SHS. Specifically, Aegis observes the states of the connected smart home entities (sensors and devices) for different user activities and usage patterns in a SHS and builds a contextual model to differentiate between malicious and benign behavior. We evaluated the efficacy and performance of Aegis in multiple smart home settings (i.e., single bedroom, double bedroom, duplex) with real-life users performing day-to-day activities and real SHS devices. We also measured the performance of Aegis against five different malicious behaviors. Our detailed evaluation shows that Aegis can detect malicious behavior in SHS with high accuracy (over 95%) and secure the SHS regardless of the smart home layout, device configuration, installed apps, and enforced user policies. Finally, Aegis achieves minimum overhead in detecting malicious behavior in SHS, ensuring easy deployability in real-life smart environments.

cs.CR↗

arXiv2019

Curie: Policy-based Secure Data Exchange

Data sharing among partners---users, organizations, companies---is crucial for the advancement of data analytics in many domains. Sharing through secure computation and differential privacy allows these partners to perform private computations on their sensitive data in controlled ways. However, in reality, there exist complex relationships among members. Politics, regulations, interest, trust, data demands and needs are one of the many reasons. Thus, there is a need for a mechanism to meet these conflicting relationships on data sharing. This paper presents Curie, an approach to exchange data among members whose membership has complex relationships. The CPL policy language that allows members to define the specifications of data exchange requirements is introduced. Members (partners) assert who and what to exchange through their local policies and negotiate a global sharing agreement. The agreement is implemented in a multi-party computation that guarantees sharing among members will comply with the policy as negotiated. The use of Curie is validated through an example of a health care application built on recently introduced secure multi-party computation and differential privacy frameworks, and policy and performance trade-offs are explored.

cs.CR↗

arXiv2018

U-PoT: A Honeypot Framework for UPnP-Based IoT Devices

The ubiquitous nature of the IoT devices has brought serious security implications to its users. A lot of consumer IoT devices have little to no security implementation at all, thus risking user's privacy and making them target of mass cyber-attacks. Indeed, recent outbreak of Mirai botnet and its variants have already proved the lack of security on the IoT world. Hence, it is important to understand the security issues and attack vectors in the IoT domain. Though significant research has been done to secure traditional computing systems, little focus was given to the IoT realm. In this work, we reduce this gap by developing a honeypot framework for IoT devices. Specifically, we introduce U-PoT: a novel honeypot framework for capturing attacks on IoT devices that use Universal Plug and Play (UPnP) protocol. A myriad of smart home devices including smart switches, smart bulbs, surveillance cameras, smart hubs, etc. uses the UPnP protocol. Indeed, a simple search on Shodan IoT search engine lists 1,676,591 UPnP devices that are exposed to public network. The popularity and ubiquitous nature of UPnP-based IoT device necessitates a full-fledged IoT honeypot system for UPnP devices. Our novel framework automatically creates a honeypot from UPnP device description documents and is extendable to any device types or vendors that use UPnP for communication. To the best of our knowledge, this is the first work towards a flexible and configurable honeypot framework for UPnP-based IoT devices. We released U-PoT under an open source license for further research and created a database of UPnP device descriptions. We also evaluated our framework on two emulated deices. Our experiments show that the emulated devices are able to mimic the behavior of a real IoT device and trick vendor-provided device management applications or popular IoT search engines while having minimal performance ovherhead.

cs.CR↗

arXiv2018

Block4Forensic: An Integrated Lightweight Blockchain Framework for Forensics Applications of Connected Vehicles

Today's vehicles are becoming cyber-physical systems that do not only communicate with other vehicles but also gather various information from hundreds of sensors within them. These developments help create smart and connected (e.g., self-driving) vehicles that will introduce significant information to drivers, manufacturers, insurance companies and maintenance service providers for various applications. One such application that is becoming crucial with the introduction of self-driving cars is the forensic analysis for traffic accidents. The utilization of vehicle-related data can be instrumental in post-accident scenarios to find out the faulty party, particularly for self-driving vehicles. With the opportunity of being able to access various information on the cars, we propose a permissioned blockchain framework among the various elements involved to manage the collected vehicle-related data. Specifically, we first integrate Vehicular Public Key Management (VPKI) to the proposed blockchain to provide membership establishment and privacy. Next, we design a fragmented ledger that will store detailed data related to vehicle such as maintenance information/history, car diagnosis reports, etc. The proposed forensic framework enables trustless, traceable and privacy-aware post-accident analysis with minimal storage and processing overhead.

cs.CR↗

arXiv2018

Detection of Compromised Smart Grid Devices with Machine Learning and Convolution Techniques

The smart grid concept has transformed the traditional power grid into a massive cyber-physical system that depends on advanced two-way communication infrastructure to integrate a myriad of different smart devices. While the introduction of the cyber component has made the grid much more flexible and efficient with so many smart devices, it also broadened the attack surface of the power grid. Particularly, compromised devices pose a great danger to the healthy operations of the smart-grid. For instance, the attackers can control the devices to change the behaviour of the grid and can impact the measurements. In this paper, to detect such misbehaving malicious smart grid devices, we propose a machine learning and convolution-based classification framework. Our framework specifically utilizes system and library call lists at the kernel level of the operating system on both resource-limited and resource-rich smart grid devices such as RTUs, PLCs, PMUs, and IEDs. Focusing on the types and other valuable features extracted from the system calls, the framework can successfully identify malicious smart-grid devices. In order to test the efficacy of the proposed framework, we built a representative testbed conforming to the IEC-61850 protocol suite and evaluated its performance with different system calls. The proposed framework in different evaluation scenarios yields very high accuracy (avg. 91%) which reveals that the framework is effective to overcome compromised smart grid devices problem.

cs.CR↗

arXiv2018

WACA: Wearable-Assisted Continuous Authentication

One-time login process in conventional authentication systems does not guarantee that the identified user is the actual user throughout the session. However, it is necessary to re-verify the user identity periodically throughout a login session without reducing the user convenience. Continuous authentication can address this issue. However, existing methods are either not reliable or not usable. In this paper, we introduce a usable and reliable method called Wearable Assisted Continuous Authentication (WACA). WACA relies on the sensor based keystroke dynamics, where the authentication data is acquired through the built in sensors of a wearable (e.g., smartwatch) while the user is typing. We implemented the WACA framework and evaluated its performance on real devices with real users. The empirical evaluation of WACA reveals that WACA is feasible and its error rate is as low as 1 percent with 30 seconds of processing time and 2 3% for 20 seconds. The computational overhead is minimal. Furthermore, we tested WACA against different attack scenarios. WACA is capable of identifying insider threats with very high accuracy (99.2%) and also robust against powerful adversaries such as imitation and statistical attackers.

cs.CR↗

arXiv2018

Sensitive Information Tracking in Commodity IoT

Broadly defined as the Internet of Things (IoT), the growth of commodity devices that integrate physical processes with digital connectivity has had profound effects on society--smart homes, personal monitoring devices, enhanced manufacturing and other IoT apps have changed the way we live, play, and work. Yet extant IoT platforms provide few means of evaluating the use (and potential avenues for misuse) of sensitive information. Thus, consumers and organizations have little information to assess the security and privacy risks these devices present. In this paper, we present SainT, a static taint analysis tool for IoT applications. SainT operates in three phases; (a) translation of platform-specific IoT source code into an intermediate representation (IR), (b) identifying sensitive sources and sinks, and (c) performing static analysis to identify sensitive data flows. We evaluate SainT on 230 SmartThings market apps and find 138 (60%) include sensitive data flows. In addition, we demonstrate SainT on IoTBench, a novel open-source test suite containing 19 apps with 27 unique data leaks. Through this effort, we introduce a rigorously grounded framework for evaluating the use of sensitive information in IoT apps---and therein provide developers, markets, and consumers a means of identifying potential threats to security and privacy.

cs.CR↗

arXiv2018

A Survey on Sensor-based Threats to Internet-of-Things (IoT) Devices and Applications

The concept of Internet of Things (IoT) has become more popular in the modern era of technology than ever before. From small household devices to large industrial machines, the vision of IoT has made it possible to connect the devices with the physical world around them. This increasing popularity has also made the IoT devices and applications in the center of attention among attackers. Already, several types of malicious activities exist that attempt to compromise the security and privacy of the IoT devices. One interesting emerging threat vector is the attacks that abuse the use of sensors on IoT devices. IoT devices are vulnerable to sensor-based threats due to the lack of proper security measurements available to control use of sensors by apps. By exploiting the sensors (e.g., accelerometer, gyroscope, microphone, light sensor, etc.) on an IoT device, attackers can extract information from the device, transfer malware to a device, or trigger a malicious activity to compromise the device. In this survey, we explore various threats targeting IoT devices and discuss how their sensors can be abused for malicious purposes. Specifically, we present a detailed survey about existing sensor-based threats to IoT devices and countermeasures that are developed specifically to secure the sensors of IoT devices. Furthermore, we discuss security and privacy issues of IoT devices in the context of sensor-based threats and conclude with future research directions.

cs.CR↗

arXiv2018

Advertising in the IoT Era: Vision and Challenges

The Internet of Things (IoT) extends the idea of interconnecting computers to a plethora of different devices, collectively referred to as smart devices. These are physical items - i.e., "things" - such as wearable devices, home appliances, and vehicles, enriched with computational and networking capabilities. Due to the huge set of devices involved - and therefore, its pervasiveness - IoT is a great platform to leverage for building new applications and services or extending existing ones. In this regard, expanding online advertising into the IoT realm is an under-investigated yet promising research direction, especially considering that traditional Internet advertising market is already worth hundreds of billions of dollars. In this paper, we first propose the architecture of an IoT advertising platform inspired by the well-known business ecosystem, which the traditional Internet advertising is based on. Additionally, we discuss the key challenges to implement such a platform with a special focus on issues related to architecture, advertisement content delivery, security, and privacy of the users.

cs.CY↗