arXiv subjects

Inaki Esnaola

Publications and source records attributed to Inaki Esnaola.

11 recordsLinked to original sources

arXiv2026

Critical Challenges and Guidelines in Evaluating Synthetic Tabular Data: A Systematic Review

Generating synthetic tabular health data is challenging, and evaluating their quality is equally, if not more, complex. This systematic review highlights the critical importance of rigorous evaluation of synthetic health data to ensure reliability, clinical relevance, and appropriate use. From an initial identification of 2067 relevant papers published in the last ten years, 134 studies were selected for detailed analysis. Our review identifies key challenges, including lack of consensus on evaluation methods, inconsistent application of evaluation metrics, limited involvement of domain experts, inadequate reporting of dataset characteristics, and limited reproducibility of results. In response, we provide a structured consolidation of synthetic data generation and evaluation methods into taxonomies, alongside practical guidelines to support more robust and standardised evaluation practices. These findings aim to support the responsible development and use of synthetic health data, aligned with emerging expectations around transparency, reproducibility, and governance, ultimately enabling the community to fully harness its transformative potential and accelerate innovation.

cs.LG↗

arXiv2025

Decentralized Stealth Attacks on Cyber-Physical Systems

Decentralized stealth attack constructions that minimize the mutual information between the state variables and the measurements are proposed. The attack constructions are formulated as random Gaussian attacks targeting Cyber-physical systems that aims at minimizing the mutual information between the state variables and measurements while constraining the Kullback-Leibler divergence between the distribution of the measurements under attacks and the distribution of the measurements without attacks. The proposed information metrics adopted measure the disruption and attack detection both globally and locally. The decentralized attack constructions are formulated in a framework of normal games. The global and local information metrics yield games with global and local objectives in disruption and attack detection. We have proven the games are potential games and the convexity of the potential functions followed by the uniqueness and the achievability of the Nash Equilibrium, accordingly. We proposed a best response dynamics to achieve the Nash Equilibrium of the games. We numerically evaluate the performance of the proposed decentralized stealth random attacks on IEEE test systems and show it is feasible to exploit game theoretic techniques in decentralized attack constructions.

eess.SY↗

arXiv2024

Submodularity of Mutual Information for Multivariate Gaussian Sources with Additive Noise

Sensor placement approaches in networks often involve using information-theoretic measures such as entropy and mutual information. We prove that mutual information abides by submodularity and is non-decreasing when considering the mutual information between the states of the network and a subset of $k$ nodes subjected to additive white Gaussian noise. We prove this under the assumption that the states follow a non-degenerate multivariate Gaussian distribution.

cs.IT↗

arXiv2019

Optimal Model-Based Sensor Placement & Adaptive Monitoring Of An Oil Spill

This paper presents a model based adaptive monitoring method for the estimation of flow tracers, with application to mapping, prediction and observation of oil spills in the immediate aftermath of an incident. Autonomous agents are guided to optimal sensing locations via the solution of a PDE constrained optimisation problem, obtained using the adjoint method. The proposed method employs a dynamic model of the combined ocean and oil dynamics, with states that are updated in real-time using a Kalman filter that fuses agent-based measurements with a reduced-order model of the ocean circulation dynamics. In turn, the updated predictions from the fluid model are used to identify and update the reduced order model, in a process of continuous feedback. The proposed method exhibits a 30% oil presence mapping and prediction improvement compared to standard industrial oil observation sensor guidance and model use.

eess.SY↗

arXiv2019

A combined ocean and oil model for model-based adaptive monitoring

This paper presents a combined ocean and oil model for adaptive placement of sensors in the immediate aftermath of oilspills. A key feature of this model is the ability to correct its predictions of spill location using continual measurement feedback from a low number of deployed sensors. This allows for a model of relatively low complexity compared to existing models, which in turn enables fast predictions. The focus of this paper is upon the modelling aspects and in-particular the trade-off between complexity and numerical efficiency. The presented model contains relevant ocean, wind and wave dynamics for short-term spill predictions. The model is used to simulate the 2019 Grande America spill, with results compared to satellite imagery. The predictions show good agreement, even after several days from the initial incident. As a precursor to future work, results are also presented that demonstrate how sensor feedback mitigates the effects of model inaccuracy.

eess.SY↗

arXiv2017

Information-Theoretic Attacks in the Smart Grid

Gaussian random attacks that jointly minimize the amount of information obtained by the operator from the grid and the probability of attack detection are presented. The construction of the attack is posed as an optimization problem with a utility function that captures two effects: firstly, minimizing the mutual information between the measurements and the state variables; secondly, minimizing the probability of attack detection via the Kullback-Leibler divergence between the distribution of the measurements with an attack and the distribution of the measurements without an attack. Additionally, a lower bound on the utility function achieved by the attacks constructed with imperfect knowledge of the second order statistics of the state variables is obtained. The performance of the attack construction using the sample covariance matrix of the state variables is numerically evaluated. The above results are tested in the IEEE 30-Bus test system.

cs.IT↗

arXiv2017

Smart Meter Privacy via the Trapdoor Channel

A battery charging policy that provides privacy guarantees for smart meter systems with finite capacity battery is proposed. For this policy an upper bound on the information leakage rate is provided. The upper bound applies for general random processes modelling the energy consumption of the user. It is shown that the average energy consumption of the user determines the information leakage rate to the utility provider. The upper bound is shown to be tight by deriving the probability law of a random process achieving the bound.

cs.IT↗

arXiv2016

Compressed Sensing Algorithms for OFDM Channel Estimation

Radio channels are typically sparse in the delay domain, and ideal for compressed sensing. A new compressed sensing algorithm called eX-OMP is developed that yields performance similar to that of the optimal MMSE estimator. The new algorithm relies on a small amount additional data. Both eX-OMP and the MMSE estimator adaptively balance channel tracking and noise reduction. They perform better than simple estimators such as the linear-interpolator which fix this trade-off a priori. Some wideband measurements are examined, and the channels are found to be represented by a few delays.

cs.IT↗

arXiv2015

Machine Learning Methods for Attack Detection in the Smart Grid

Attack detection problems in the smart grid are posed as statistical learning problems for different attack scenarios in which the measurements are observed in batch or online settings. In this approach, machine learning algorithms are used to classify measurements as being either secure or attacked. An attack detection framework is provided to exploit any available prior knowledge about the system and surmount constraints arising from the sparse structure of the problem in the proposed approach. Well-known batch and online learning algorithms (supervised and semi-supervised) are employed with decision and feature level fusion to model the attack detection problem. The relationships between statistical and geometric properties of attack vectors employed in the attack scenarios and learning algorithms are analyzed to detect unobservable attacks using statistical learning methods. The proposed algorithms are examined on various IEEE test systems. Experimental analyses show that machine learning algorithms can detect attacks with performances higher than the attack detection algorithms which employ state vector estimation methods in the proposed attack detection framework.

cs.LG↗

arXiv2015

Sparse Attack Construction and State Estimation in the Smart Grid: Centralized and Distributed Models

New methods that exploit sparse structures arising in smart grid networks are proposed for the state estimation problem when data injection attacks are present. First, construction strategies for unobservable sparse data injection attacks on power grids are proposed for an attacker with access to all network information and nodes. Specifically, novel formulations for the optimization problem that provide a flexible design of the trade-off between performance and false alarm are proposed. In addition, the centralized case is extended to a distributed framework for both the estimation and attack problems. Different distributed scenarios are proposed depending on assumptions that lead to the spreading of the resources, network nodes and players. Consequently, for each of the presented frameworks a corresponding optimization problem is introduced jointly with an algorithm to solve it. The validity of the presented procedures in real settings is studied through extensive simulations in the IEEE test systems.

cs.IT↗

arXiv2013

Power Allocation Strategies in Energy Harvesting Wireless Cooperative Networks

In this paper, a wireless cooperative network is considered, in which multiple source-destination pairs communicate with each other via an energy harvesting relay. The focus of this paper is on the relay's strategies to distribute the harvested energy among the multiple users and their impact on the system performance. Specifically, a non-cooperative strategy is to use the energy harvested from the i-th source as the relay transmission power to the i-th destination, to which asymptotic results show that its outage performance decays as logSNR over SNR. A faster decaying rate, 1 over SNR, can be achieved by the two centralized strategies proposed this the paper, where the water filling based one can achieve optimal performance with respect to several criteria, with a price of high complexity. An auction based power allocation scheme is also proposed to achieve a better tradeoff between the system performance and complexity. Simulation results are provided to confirm the accuracy of the developed analytical results and facilitate a better performance comparison.

cs.IT↗