CyberSec Research

Over-the-air (OTA) federated learning (FL) has been well recognized as a scalable paradigm that exploits the waveform superposition of the wireless multiple-access channel to aggregate model updates in a single use. Existing OTA-FL designs largely enforce zero-bias model updates by either assuming \emph{homogeneous} wireless conditions (equal path loss across devices) or forcing zero-bias updates to guarantee convergence. Under \emph{heterogeneous} wireless scenarios, however, such designs are constrained by the weakest device and inflate the update variance. Moreover, prior analyses of biased OTA-FL largely address convex objectives, while most modern AI models are highly non-convex. Motivated by these gaps, we study OTA-FL with stochastic gradient descent (SGD) for general smooth non-convex objectives under wireless heterogeneity. We develop novel OTA-FL SGD updates that allow a structured, time-invariant model bias while facilitating reduced variance updates. We derive a finite-time stationarity bound (expected time average squared gradient norm) that explicitly reveals a bias-variance trade-off. To optimize this trade-off, we pose a non-convex joint OTA power-control design and develop an efficient successive convex approximation (SCA) algorithm that requires only statistical CSI at the base station. Experiments on a non-convex image classification task validate the approach: the SCA-based design accelerates convergence via an optimized bias and improves generalization over prior OTA-FL baselines.

This paper presents a time-optimal Model Predictive Control (MPC) scheme for linear discrete-time systems subject to multiplicative uncertainties represented by interval matrices. To render the uncertainty propagation computationally tractable, the set-valued error system dynamics are approximated using a matrix-zonotope-based bounding operator. Recursive feasibility and finite-time convergence are ensured through an adaptive terminal constraint mechanism. A key advantage of the proposed approach is that all the necessary bounding sets can be computed offline, substantially reducing the online computational burden. The effectiveness of the method is illustrated via a numerical case study on an orbital rendezvous maneuver between two satellites.

Guaranteeing stringent data freshness for low-altitude unmanned aerial vehicles (UAVs) in shared spectrum forces a critical trade-off between two operational costs: the UAV's own energy consumption and the occupation of terrestrial channel resources. The core challenge is to satisfy the aerial data freshness while finding a Pareto-optimal balance between these costs. Leveraging predictive channel models and predictive UAV trajectories, we formulate a bi-objective Pareto optimization problem over a long-term planning horizon to jointly optimize the sampling timing for aerial traffic and the power and spectrum allocation for fair coexistence. However, the problem's non-convex, mixed-integer nature renders classical methods incapable of fully characterizing the complete Pareto frontier. Notably, we show monotonicity properties of the frontier, building on which we transform the bi-objective problem into several single-objective problems. We then propose a new graph-based algorithm and prove that it can find the complete set of Pareto optima with low complexity, linear in the horizon and near-quadratic in the resource block (RB) budget. Numerical comparisons show that our approach meets the stringent timeliness requirement and achieves a six-fold reduction in RB utilization or a 6 dB energy saving compared to benchmarks.

The proposed approach yields a numerical method that provably executes in linear time with respect to the number of nodes and edges in a graph. The graph, constructed from the power system model, requires only knowledge of the dependencies between state-to-state and output-to-state variables within a state-space framework. While graph-based observability analysis methods exist for power system static-state estimation, the approach presented here is the first for dynamic-state estimation (DSE). We examine decentralized and centralized DSE scenarios and compare our findings with a well-established, albeit non-scalable, observability analysis method in the literature. When compared to the latter in a centralized DSE setting, our method reduced computation time by 1440x.

The proliferation of inverter-based resources challenges traditional microgrid protection by introducing variable fault currents and complex transients. This paper presents a statistically adaptive differential protection scheme based on Kullback-Leibler divergence, implemented via a Bartlett-corrected G-statistic computed on logarithm-transformed current magnitudes. The method is a multivariate fault detection engine that employs the Mahalanobis distance to distinguish healthy and faulty states, enabling robust detection even in noisy environments. Detection thresholds are statistically derived from a chi-squared distribution for precise control over the false alarm rate. Upon detection, a lightweight classifier identifies the fault type by assessing per-phase G-statistics against dedicated thresholds, enhanced by a temporal persistence filter for security. Extensive simulations on a modified CIGRE 14-bus microgrid show high efficacy: sub-cycle average detection delays, high detection and classification accuracy across operating modes, resilience to high-impedance faults up to 250 Ohms, tolerance to 10 ms communication delay, and noise levels down to a 20 dB signal-to-noise ratio. These findings demonstrate a reproducible and computationally efficient solution for next-generation AC microgrid protection.

The electricity sector transition requires substantial increases in residential demand response capacity, yet Home Energy Management Systems (HEMS) adoption remains limited by user interaction barriers requiring translation of everyday preferences into technical parameters. While large language models have been applied to energy systems as code generators and parameter extractors, no existing implementation deploys LLMs as autonomous coordinators managing the complete workflow from natural language input to multi-appliance scheduling. This paper presents an agentic AI HEMS where LLMs autonomously coordinate multi-appliance scheduling from natural language requests to device control, achieving optimal scheduling without example demonstrations. A hierarchical architecture combining one orchestrator with three specialist agents uses the ReAct pattern for iterative reasoning, enabling dynamic coordination without hardcoded workflows while integrating Google Calendar for context-aware deadline extraction. Evaluation across three open-source models using real Austrian day-ahead electricity prices reveals substantial capability differences. Llama-3.3-70B successfully coordinates all appliances across all scenarios to match cost-optimal benchmarks computed via mixed-integer linear programming, while other models achieve perfect single-appliance performance but struggle to coordinate all appliances simultaneously. Progressive prompt engineering experiments demonstrate that analytical query handling without explicit guidance remains unreliable despite models' general reasoning capabilities. We open-source the complete system including orchestration logic, agent prompts, tools, and web interfaces to enable reproducibility, extension, and future research.

The increasing integration of renewable energy sources and distributed energy resources (DER) into modern power systems introduces significant uncertainty, posing challenges for maintaining grid flexibility and reliability. Hybrid energy systems (HES), composed of controllable generators, flexible loads, and battery storage, offer a decentralized solution to enhance flexibility compared to single centralized resources. This paper presents a two-level framework to enable HES participation in frequency regulation markets. The upper level performs a chance-constrained optimization to choose capacity bids based on historical regulation signals. At the lower level, a real-time control strategy disaggregates the regulation power among the constituent resources. This real-time control strategy is then benchmarked against an offline optimal dispatch to evaluate flexibility performance. Additionally, the framework evaluates the profitability of overbidding strategies and identifies thresholds beyond which performance degradation may lead to market penalties or disqualification. The proposed framework also compare the impact of imbalance of power capacities on performance and battery state of charge (SoC) through asymmetric HES configurations.

In post-disaster scenarios, the rapid deployment of adequate communication infrastructure is essential to support disaster search, rescue, and recovery operations. To achieve this, uncrewed aerial vehicle (UAV) has emerged as a promising solution for emergency communication due to its low cost and deployment flexibility. However, conventional untethered UAV (U-UAV) is constrained by size, weight, and power (SWaP) limitations, making it incapable of maintaining the operation of a macro base station. To address this limitation, we propose a heterogeneous UAV-based framework that integrates tethered UAV (T-UAV) and U-UAVs, where U-UAVs are utilized to enhance the throughput of cell-edge ground user equipments (G-UEs) and guarantee seamless connectivity during G-UEs' mobility to safe zones. It is noted that the integrated access and backhaul (IAB) technique is adopted to support the wireless backhaul of U-UAVs. Accordingly, we formulate a two-timescale joint user scheduling and trajectory control optimization problem, aiming to maximize the downlink throughput under asymmetric traffic demands and G-UEs' mobility. To solve the formulated problem, we proposed a two-timescale multi-agent deep deterministic policy gradient (TTS-MADDPG) algorithm based on the centralized training and distributed execution paradigm. Numerical results show that the proposed algorithm outperforms other benchmarks, including the two-timescale multi-agent proximal policy optimization (TTS-MAPPO) algorithm and MADDPG scheduling method, with robust and higher throughput. Specifically, the proposed algorithm obtains up to 12.2\% average throughput gain compared to the MADDPG scheduling method.

People tend to walk in groups, and interactions with those groups have a significant impact on crowd behavior and pedestrian traffic dynamics. Social norms can be seen as unwritten rules regulating people interactions in social settings. This article studies people interactions with groups and the emergence of group proxemics. Group zones, zone occupancy counts and people clearance from the group are studied using naturalistic data. Analysis indicate potential presence of three different zones in addition to the public zone. People tend to remain in the public zone and only progressively get closer to groups, and those closer approaches happen in a low frequency and for brief periods of time.

This article investigates the pedestrian group as an emergent agent. The article explores empirical data to derive emergent agency and formation state spaces and outline recurring patterns of walking behavior. In this analysis, pedestrian trajectories extracted from surveillance videos are used along with manually annotated pedestrian group memberships. We conducted manual expert evaluation of observed groups, produced new manual annotations for relevant events pertaining to group behavior and extracted metrics relevant group formation. This information along with quantitative analysis was used to model the life-cycle and formation of the group agent. Those models give structure to expectations around walking behavior of groups; from pedestrian walking independently to the emergence of a collective intention where group members tended to maintain bounded distance between each other. Disturbances to this bounded distance often happened in association with changes in either their agency or their formation states. We summarized the patterns of behavior along with the sequences of state transitions into abstract patterns, which can aid in the development of more detailed group agents in simulation and in the design of engineering systems to interact with such groups.

This article proposes a modular optimal control framework for local three-dimensional ellipsoidal obstacle avoidance, exemplarily applied to model predictive path-following control. Static as well as moving obstacles are considered. Central to the approach is a computationally efficient and continuously differentiable condition for detecting collisions with ellipsoidal obstacles. A novel two-stage optimization approach mitigates numerical issues arising from the structure of the resulting optimal control problem. The effectiveness of the approach is demonstrated through simulations and real-world experiments with the Crazyflie quadrotor. This represents the first hardware demonstration of an MPC controller of this kind for UAVs in a three-dimensional task.

We investigate the gain margin of a general nonlinear system under an inverse optimal input-to-state safe (ISSf) controller of the form u=u0(x)+u*(x,u0), where u0 is the nominal control and u* is the inverse optimal safety filter that minimally modifies the nominal controller's unsafe actions over the infinite horizon. By first establishing a converse ISSf-BF theorem, we reveal the equivalence among the achievability of ISSf by feedback, the achievability of inverse optimality, and the solvability of a Hamilton-Jacobi-Isaacs equation associated with the inverse optimal ISSf gain assignment. Then we develop a collection of safety margin results on the overall control u=u0+u*. In the absence of disturbances, we find that standard inverse optimal safe controllers have a certain degree of gain margin. Specifically, when f(x) acts safely but u0 acts unsafely, the gain can be decreased by up to half; and when f(x) acts unsafely, we establish that, if u0 acts safely, the gain can be increased arbitrarily, whereas if u0 acts unsafely, the control recovers the full gain margin [1/2,inf). It is shown, however, that under control gain variation, the safe set of these controllers is locally asymptotically stable, which implies that their safety is sensitive to large but bounded disturbances. To make inverse optimal ISSf controllers robust to gain variation, we propose a gain margin improvement approach at the expense of an increased control effort. This improvement allows the inverse optimal safe control to inherit the standard gain margin of [1/2,inf) without requiring prior knowledge of whether f(x) or u0 acts safely on the safety boundary, while simultaneously ensuring global asymptotic stability of the resulting safe set. In the presence of disturbances, this improvement idea renders inverse optimal ISSf controllers robust to gain variations with the same gain margin of [1/2,inf).

The manufacturing sector is moving from rigid, hardware-dependent systems toward flexible, software-driven environments. This transformation is shaped by the convergence of several Software-Defined technologies: Software-Defined Automation virtualizes industrial control, replacing proprietary PLCs with containerized, programmable solutions that enable scalability and interoperability. Software-Defined Compute and Communications provide a means to distribute intelligence seamlessly across devices, networks, and cloud platforms, reducing latency and enabling dynamic reconfiguration. Software-Defined Manufacturing Systems, usually implemented as Digital Twins, are real-time virtual models of machines and processes, allowing predictive analysis, optimization, and closer integration between human operators and intelligent systems. This work presents XWAVE, a project that unites these three Software-Defined paradigms to present a modular, fully software-defined manufacturing system.

We propose a command-filter backstepping controller that integrates a disturbance observer and a high-gain observer (HGO) to handle unknown internal and external disturbances acting on a quadrotor. To build the controller, we first define tracking errors between the measured and desired quadrotor outputs, which allow the system to be rewritten in a new set of state variables. Using this transformed model, we apply Lyapunov theory to derive a backstepping control law. To avoid repeated differentiation of states and virtual controls, a first-order command filter is introduced, and a nonlinear disturbance observer is added to provide disturbance estimates. Each state in the controller and observer is replaced with its estimate from the HGO. The resulting control law enables the quadrotor to follow its path despite internal and external disturbances, with each subsystem allowed its own disturbance type for realism. A new state transformation and Lyapunov-based derivation prevent the usual explosion of complexity, while the HGO reconstructs unmeasured states and their rates for output feedback. The nonlinear disturbance observer attenuates constant and nonlinear disturbances as well as band-limited white noise. The method reduces dependence on high-precision sensors and mitigates wind, model error, and rotor noise effects during flight. Unlike previous studies that treat either disturbance rejection or partial sensing, this work combines the command filter, disturbance observer, and HGO to address both challenges simultaneously while avoiding the complexity growth typical of backstepping designs.

Many tasks in human environments require collaborative behavior between multiple kinematic chains, either to provide additional support for carrying big and bulky objects or to enable the dexterity that is required for in-hand manipulation. Since these complex systems often have a very high number of degrees of freedom coordinating their movements is notoriously difficult to model. In this article, we present the derivation of the theoretical foundations for cooperative task spaces of multi-arm robotic systems based on geometric primitives defined using conformal geometric algebra. Based on the similarity transformations of these cooperative geometric primitives, we derive an abstraction of complex robotic systems that enables representing these systems in a way that directly corresponds to single-arm systems. By deriving the associated analytic and geometric Jacobian matrices, we then show the straightforward integration of our approach into classical control techniques rooted in operational space control. We demonstrate this using bimanual manipulators, humanoids and multi-fingered hands in optimal control experiments for reaching desired geometric primitives and in teleoperation experiments using differential kinematics control. We then discuss how the geometric primitives naturally embed nullspace structures into the controllers that can be exploited for introducing secondary control objectives. This work, represents the theoretical foundations of this cooperative manipulation control framework, and thus the experiments are presented in an abstract way, while giving pointers towards potential future applications.

This paper studies how to achieve accurate modeling and effective control in stochastic nonlinear dynamics with multiple interacting objects. However, non-uniform interactions and random topologies make this task challenging. We address these challenges by proposing \textit{Graph Controllable Embeddings} (GCE), a general framework to learn stochastic multi-object dynamics for linear control. Specifically, GCE is built on Hilbert space embeddings, allowing direct embedding of probability distributions of controlled stochastic dynamics into a reproducing kernel Hilbert space (RKHS), which enables linear operations in its RKHS while retaining nonlinear expressiveness. We provide theoretical guarantees on the existence, convergence, and applicability of GCE. Notably, a mean field approximation technique is adopted to efficiently capture inter-object dependencies and achieve provably low sample complexity. By integrating graph neural networks, we construct data-dependent kernel features that are capable of adapting to dynamic interaction patterns and generalizing to even unseen topologies with only limited training instances. GCE scales seamlessly to multi-object systems of varying sizes and topologies. Leveraging the linearity of Hilbert spaces, GCE also supports simple yet effective control algorithms for synthesizing optimal sequences. Experiments on physical systems, robotics, and power grids validate GCE and demonstrate consistent performance improvement over various competitive embedding methods in both in-distribution and few-shot tests

This study analyzes the performance of positioning techniques based on configuration changes of 5G New Radio signals. In 5G networks, a terminal position is determined from the Time of Arrival of Positioning Reference Signals transmitted by base stations. We propose an algorithm that improves TOA accuracy under low sampling rate constraints and implement 5G PRS for positioning in a software defined modem. We also examine how flexible time frequency resource allocation of PRS affects TOA estimation accuracy and discuss optimal PRS configurations for a given signal environment.

Edge computing alleviates the computation burden of data-driven control in cyber-physical systems (CPSs) by offloading complex processing to edge servers. However, the increasing sophistication of cyberattacks underscores the need for security measures that go beyond conventional IT protections and address the unique vulnerabilities of CPSs. This study proposes a confidential data-driven gain-tuning framework using homomorphic encryption, such as ElGamal and CKKS encryption schemes, to enhance cybersecurity in gain-tuning processes outsourced to external servers. The idea for realizing confidential FRIT is to replace the matrix inversion operation with a vector summation form, allowing homomorphic operations to be applied. Numerical examples under 128-bit security confirm performance comparable to conventional methods while providing guidelines for selecting suitable encryption schemes for secure CPS.