Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Data privacy”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 145 records · Page 8

Extended convex hull-based distributed optimal energy flow of integrated electricity-gas systems

Integrated electricity and gas systems are constructed to facilitate the gas-fired generation, and the distributed operation of these integrated systems have received much attention due to the increased emphasis on data security and privacy between different agencies. This paper proposes an extended convex hull based method to address optimal energy flow problems for the integrated electricity and gas systems in a distributed manner. First, a multi-block electricity-gas system model is constructed by dividing the whole system into N blocks considering both physical and regional differences. This multi-block model is then convexified by replacing the nonconvex gas transmission equation with the extended convex hull-based constraints. The Jacobi-Proximal alternating direction method of multipliers algorithm is adopted to solve the convexified model and minimize its operation cost. Finally, the feasibility of the optimal solution for the convexified model is checked, and a sufficient condition is developed. If the sufficient condition is satisfied, the optimal solution for the original nonconvex problem can be recovered from that for the convexified problem. Simulation results demonstrate that the proposed method is tractable and effective in obtaining feasible optimal solutions for multi-block optimal energy flow problems.

24 POWER TRANSMISSION AND DISTRIBUTION↗

A Multi-Institutional Study of Feasibility, Implementation, and Early Clinical Results With Noninvasive Breast Brachytherapy for Tumor Bed Boost

Purpose: To evaluate the feasibility, implementation, and early results of noninvasive breast brachytherapy (NIBB) for tumor bed boost with whole breast radiation therapy (WBRT). Methods and Materials: NIBB is a commercially available (AccuBoost, Billerica, MA) mammography-based, brachytherapy system in which the treatment applicators are centered on the planning target volume (PTV) to direct {sup 192}Ir emissions along orthogonal axes. A privacy-encrypted online data registry collected information from 8 independent academic and community-based institutions. Data were from 146 consecutive women with early-stage breast cancer after lumpectomy and WBRT receiving boost with NIBB between July 2007 and March 2010. Toxicity and cosmesis were graded according to the Common Toxicity Criteria (v. 3.0) and the Harvard scale. Median follow-up was 6 months (1-39 months). Results: Grade 1-2 skin toxicity was observed in 64%, 48%, and 21% during the acute (1-3 weeks), intermediate (4-26 weeks), and late-intermediate (>26 weeks) periods. There was no Grade 4 toxicity. At 6 months, for the entire cohort, cosmesis was excellent/good in 62%/38%. The subset receiving NIBB before WBRT had cosmetic scores of 32% and 63%, whereas during WBRT, 58% and 37% were rated as excellent and good, respectively. Breast compression was scored as 'uncomfortable' in 12%, 29%, and 59% when NIBB was delivered before, during, or after WBRT. For each patient, the fraction-to-fraction variability in PTV was low. Skin flash was associated with a higher proportion of excellent cosmesis (58% vs. 42%) relative to having the applicator all within breast tissue. Conclusions: These data indicate that NIBB is feasible and can be consistently implemented in a broad array of practice settings. Preliminary evaluation suggests that NIBB is associated with acceptably mild normal tissue toxicity and favorable early cosmesis. The application of NIBB before WBRT may be associated with better patient tolerance at the expense of less favorable cosmetic outcome.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Enhancing Cyber Resilience of Networked Microgrids using Vertical Federated Reinforcement Learning

This paper presents a novel federated reinforcement learning (Fed-RL) methodology to inject sufficient resiliency into the operations of the network of microgrids. We consider adversarial actions to the voltage and power control loop reference signals at the grid forming (GFM) inverters in the microgrids which are essential to integrate renewable resources. Therefore, we formulate a resilient reinforcement learning training setup that uses these adversarial injections to generate episodic trajectories and train the RL agents to alleviate their impact on performance. To circumvent the concerns about data-sharing and privacy for different owners of the microgrids in the networked setting, we bring in the aspects of the federated operation to propose novel Fed-RL algorithms. As the dynamics of each microgrid are coupled due to electrical interlinks, the conventional federated RL approaches using decoupled independent environments are not applicable, which leads us to propose a multi-agent vertically federated variation of actor-critic algorithms, namely federated soft actor-critic (FedSAC). We have performed numerical simulations on an IEEE 123-bus benchmark test feeder with three microgrids by creating a customized simulation setup by encapsulating the microgrid dynamic simulations in GridLAB-D/HELICS co-simulation platform with the OpenAI Gym environment and validated the proposed resilient and secured learning methodology.

Artificial Intelligence (AI), reinforcement learni↗

Resilient Control of Networked Microgrids Using Vertical Federated Reinforcement Learning: Designs and Real-Time Test-Bed Validations

Improving system-level resiliency of networked microgrids against adversarial cyber-attacks is an important aspect in the current regime of increased inverter-based resources (IBRs). To achieve that, this paper contributes in designing a hierarchical control layer, in conjunction with the existing control layers, resilient to adversarial attack signals. Considering model complexities, unknown dynamical behaviors of IBRs, and privacy issues regarding data sharing in multi-party-owned microgrids, designing such a control layer is non-trivial. Here, to tackle these issues, a novel federated reinforcement learning (Fed-RL) method is proposed. To grasp the interconnected dynamics of networked microgrids, the paper develops Federated Soft Actor-Critic (FedSAC) algorithm following the vertical structure of implementing Fed-RL. Next, utilizing the OpenAI Gym interface, we built a custom set-up in GridLAB-D/HELICS co-simulation platform, named Resilient RL Co-simulation (ResRLCoSIM), to train the RL agents with IEEE 123-bus benchmark comprising 3 interconnected microgrids. Finally, the learned policies in the simulation are transferred to the real-time hardware-in-the-loop (HIL) test-bed developed using the high-fidelity Hypersim platform. Finally, experiments show that the simulator-trained RL controllers achieve desirable performance with the test-bed platform, validating the minimization of the sim-to-real gap.

24 POWER TRANSMISSION AND DISTRIBUTION↗

CyberGAN: Generating High-fidelity Cybersecurity Data With Generative Adversarial Networks

Machine learning for cyber defense offers the promise of detecting adversarial activity against the ground data systems managing critical space assets. A fundamental challenge facing machine learning research in cybersecurity is the lack of high-fidelity, shareable datasets for robust evaluation and testing of machine learning-based solutions. High-fidelity, real-world datasets are necessary for reliable benchmarking of nominal system behavior and malicious activity. Unfortunately, such realistic datasets of both nominal and adversarial activity are rarely shared publicly by data owners due to security and privacy concerns. Besides, the available adversarial data is sparse, which makes training models on malicious activity much harder. This situation has impeded and continues to impede the research and successful adoption of machine learning methods for cyber defense. Researchers have dealt with this problem by generating data within a low-fidelity lab environment, using classified and thus unshareable datasets, or downloading low-fidelity public datasets made available by others. We propose an innovative solution to the problem by employing machine learning methods to generate high-fidelity data. Specifically, we propose the use of Generative Adversarial Networks (GANs) to generate high-fidelity data for cybersecurity purposes. GANs have found successful image processing and natural language applications, but have not yet been investigated for cyber data generation. Our proposed approach first involves training the `discriminator' network of the GAN with a sample of real-world data consisting of malicious and nominal samples. We then use the `generator' network to generate new high-fidelity data samples consisting of an appropriate mix of malicious and nominal activity. We demonstrate applications of our architecture by generating high-fidelity cybersecurity data containing both malicious and nominal samples. We thoroughly evaluate the fidelity of our generated data using heuristics and evaluate its usefulness for machine learning applications using three different datasets. Overall, our approach results in high-fidelity, shareable datasets.

Zhang, Yuening↗

Challenges and Opportunities for Next-Generation Manufacturing in Space

With commercial space travel now a reality, the idea that people might spend time on other planets in the future seems to have greater potential. To make this possible, however, there needs to be flexible means for manufacturing in space to enable tooling or resources to be created when needed to handle unexpected situations. Next-generation manufacturing paradigms offer significant potential for the kind of flexibility that might be needed; however, they can result in increases in computation time compared to traditional control methods that could make many of the computing resources already available on earth attractive for use. Furthermore, resilience is a significant focus of next-generation manufacturing strategies, and one way to enable resilience for space manufacturing would be to have backup controllers available on earth. These types of considerations raise questions about remote control and monitoring, as well as privacy of the data involved in such practices, that must be considered. This work provides a perspective on several topics tied to remote control and monitoring for manufacturing in space.

Kip Nieman↗

Preliminary Routing for a National Laboratory Lab-Wide Commuting Program

There is a growing interest in reducing the amount of vehicle miles traveled (VMT), both in public and private sector organizations, in an effort to reduce overall carbon emissions. The Oak Ridge National Laboratory (ORNL), located in Oak Ridge, Tennessee, and located within close proximity to Knoxville, Tennessee, has more than 4,500 employees, most of which live in and around Knoxville. ORNL is currently developing a pilot commuting program for all employees, which incorporates the use of park-and-ride facilities. This study outlines the methodology behind the preliminary geoanalysis and routing used in developing a lab-wide commuting program. Due to privacy concerns, the data used for the study included numbers of employee residences per zip code rather than actual residential addresses. Commuting configurations by clustered zip code area and vehicle type were developed based on number of residences in a limited area, and the desire to minimize VMT. Central locations were selected using ArcGIS software. Satellite imagery was used to locate actual, suitable parking facilities to accommodate the specified number of residents involved in each commuting configuration. Optimized routing, and estimations of travel times, were performed using TransCAD GIS software. Energy estimates in kilowatt-hours (kwh) and gallons of gasoline, and gallons of gasoline equivalent, were all determined based on the scenarios. Standard petroleum-fueled vehicles were used in the initial estimates. Standard electric vehicles were also used in alternative scenarios to estimate potential additional energy and fuel savings. Furthermore the initial findings from this work will be used to develop a pilot program for ORNL.

99 GENERAL AND MISCELLANEOUS↗

Faster approximate subgraph counts with privacy

One of the most common problems studied in the context of differential privacy for graph data is counting the number of non-induced embeddings of a subgraph in a given graph. These counts have very high global sensitivity. Therefore, adding noise based on powerful alternative techniques, such as smooth sensitivity and higher-order local sensitivity have been shown to give significantly better accuracy. However, all these alternatives to global sensitivity become computationally very expensive, and to date efficient polynomial time algorithms are known only for few selected subgraphs, such as triangles, k-triangles, and k-stars. In this paper, we show that good approximations to these sensitivity metrics can be still used to get private algorithms. Using this approach, we much faster algorithms for privately counting the number of triangles in real-world social networks, which can be easily parallelized. We also give a private polynomial time algorithm for counting any constant size subgraph using less noise than the global sensitivity; we show this can be improved significantly for counting paths in special classes of graphs

Nguyen, Dung↗

Employing a Hardware-in-the-Loop Approach to Realize a Fully Homomorphic Controller for a Small Modular Advanced High Temperature Reactor

This paper addresses the cybersecurity challenges of advanced nuclear reactors by integrating fully homomorphic encryption (FHE) into their control systems, enabling encrypted processing of control signals without compromising functionality. Advanced nuclear reactors, including Small Modular Reactors (SMRs) and microreactors, aim to achieve autonomous and remote operations, reducing costs and enhancing competitiveness. However, these advancements expand the attack surface for cyberattacks, particularly in autonomous and remote operation scenarios. Cyberattacks can exploit vulnerabilities to manipulate physical processes, causing shutdowns, asset damage, or public harm. Such attacks begin with passive reconnaissance, where adversaries intercept communications or observe behaviors to gather information, which is then leveraged to execute cyber-physical attacks by injecting malicious commands. Nuclear power must adopt cybersecurity protection measures to secure the integrity and availability of their digital control systems. This paper demonstrates the application of FHE to secure operations by enabling encrypted processing of sensitive signals and parameters -- ensuring privacy without exposing data. FHE supports secure mathematical operations on encrypted data without requiring decryption. Using a hardware-in-the-loop (HIL) approach, this paper implements an FHE-integrated controller on a BeagleBone Black (BBB) controlling a simulation of the Small Modular Advanced High Temperature Reactor (SmAHTR). By doing so, the encrypted controller protects the integrity of critical set points and control signals during transmission and processing. Thus, FHE-integrated controllers enhance secure operations of advanced nuclear reactors while maintaining functionality.

control systems↗

Adversarial Training for Privacy-Preserving Deep Learning Model Distribution

Collaboration among cancer registries is essential to develop accurate, robust, and generalizable deep learning models for automated information extraction from cancer pathology reports. Sharing data presents a serious privacy issue, especially in biomedical research and healthcare delivery domains. Distributing pretrained deep learning (DL) models has been proposed to avoid critical data sharing. However, there is growing recognition that collaboration among clinical institutes through DL model distribution exposes new security and privacy vulnerabilities. These vulnerabilities increase in natural language processing (NLP) applications, in which the dataset vocabulary with word vector representations needs to be associated with the other model parameters. In this paper, we propose a novel privacy-preserving DL model distribution across cancer registries for information extraction from cancer pathology reports with privacy and confidentiality considerations. The proposed approach exploits the adversarial training framework to distinguish private features from shared features among different datasets. It only shares registry-invariant model parameters, without sharing raw data nor registry-specific model parameters among cancer registries. Thus, it protects both the data and the trained model simultaneously. We compare our proposed approach to single-registry models, and a model trained on centrally hosted data from different cancer registries. The results show that the proposed approach significantly outperforms the single-registry models and achieves statistically indistinguishable micro and macro F1-score as compared to the centralized model.

Alawad, Mohammed↗

Establishing a Secure Data Center with Remote Access: Preprint

Access to existing travel data is critical for many analysis efforts that lack the time or resources to support detailed data collection. High-resolution data sets provide particular value, but also present a challenge for preserving the anonymity of the original survey participants. To address this dilemma of providing data access while preserving privacy, the National Renewable EnergyLaboratory and the U.S. Department of Transportation have launched the Transportation Secure Data Center (TSDC). TSDC data sets include those from regional travel surveys and studies that increasingly use global positioning system devices. Data provided by different collecting agencies varies with respect to formatting, elements included and level of processing conducted in support of theoriginal purpose. The TSDC relies on a number of geospatial and other analysis tools to ensure data quality and to generate useful information outputs. TSDC users can access the processed data in two different ways. The first is by downloading summary results and second-by-second vehicle speed profiles (with latitude/longitude information removed) from a publicly-accessible website. The secondmethod involves applying for a remote connection account to a controlled-access environment where spatial analysis can be conducted, but raw data cannot be removed. The TSDC website is http://www.nrel.gov/vehiclesandfuels/secure_transportation_data.html.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Development and Preliminary Analysis of a U.S. Geothermal Heat Pump Installation Database

This paper seeks to addresses the significant gap in the literature regarding the installation and adoption of geothermal heat pump (GHP) systems in the United States. While the "2021 U.S. Geothermal Power Production and District Heating Market Report" published by the National Renewable Energy Laboratory (NREL) focused on direct-use geothermal district heating systems, it did not include an analysis of GHP installations (Robins et al. 2021). To bridge this gap, NREL has compiled a novel database currently containing 70,470 records of GHP installations, primarily sourced from state well permits and small-scale studies. Our methodology emphasizes the collection, cleaning, and standardization of data, addressing challenges such as inconsistent reporting formats and privacy concerns. Despite limitations in data on capacity, costs, and performance, our preliminary geospatial analysis reveals insights into the distribution of GHP systems across urban and rural areas and climate zones. The paper highlights the importance of publicly accessible data for advancing GHP technology adoption with a discussion of existing data sources and their limitations, advocating for improved collaboration between NREL and industry stakeholders.

data collection↗

The price of privately releasing contingency tables, and the spectra of random matrices with correlated rows

Contingency tables are the method of choice of government agencies for releasing statistical summaries of categorical data. In this paper, we consider lower bounds on how much distortion (noise) is necessary in these tables to provide privacy guarantees when the data being summarized is sensitive. We extend a line of recent work on lower bounds on noise for private data analysis [10, 13. 14, 15] to a natural and important class of functionalities. Our investigation also leads to new results on the spectra of random matrices with correlated rows. Consider a database D consisting of n rows (one per individual), each row comprising d binary attributes. For any subset of T attributes of size |T| = k, the marginal table for T has 2{sup k} entries; each entry counts how many times in the database a particular setting of these attributes occurs. Imagine an agency that wishes to release all (d/k) contingency tables for a given database. For constant k, previous work showed that distortion {tilde {Omicron}}(min{l_brace}n, (n{sup 2}d){sup 1/3}, {radical}d{sup k}{r_brace}) is sufficient for satisfying differential privacy, a rigorous definition of privacy that has received extensive recent study. Our main contributions are: (1) For {epsilon}- and ({epsilon}, {delta})-differential privacy (with {epsilon} constant and {delta} = 1/poly(n)), we give a lower bound of {tilde {Omega}}(min{l_brace}{radical}n, {radical}d{sup k}{r_brace}), which is tight for n = {tilde {Omega}}(d{sup k}). Moreover, for a natural and popular class of mechanisms based on additive noise, our bound can be strengthened to {Omega}({radical}d{sup k}), which is tight for all n. Our bounds extend even to non-constant k, losing roughly a factor of {radical}2{sup k} compared to the best known upper bounds for large n. (2) We give efficient polynomial time attacks which allow an adversary to reconstruct sensitive infonnation given insufficiently perturbed contingency table releases. For constant k, we obtain a lower bound of {tilde {Omega}}(min{l_brace}{radical}n, {radical}d{sup k}{r_brace}) that applies to a large class of privacy notions, including K-anonymity (along with its variants) and differential privacy. In contrast to our bounds for differential privacy, this bound (a) is shown only for constant k, but (b) is tight for all values of n when k is constant. (3) Our reconstruction-based attacks require a new lower bound on the least singular values of random matrices with correlated rows. For a constant k, consider a matrix M with (d/k) rows which are formed by taking all possible k-way entry-wise products of an underlying set of d random vectors. We show that even for nearly square matrices with d{sup k}/log d columns, the least singular value is {Omega}({radical}d{sup k}) with high probability - asymptotically, the same bound as one gets for a matrix with independent rows. The proof requires several new ideas for analyzing random matrices and could be of independent interest.

97 MATHEMATICS AND COMPUTING↗

Scale-up Unlearnable Examples Learning with High-performance Computing

Recent advancements in AI models, like ChatGPT, are structured to retain user interactions, which could inadvertently include sensitive healthcare data. In the healthcare field, particularly when radiologists use AI-driven diagnostic tools hosted on online platforms, there is a risk that medical imaging data may be repurposed for future AI training without explicit consent, spotlighting critical privacy and intellectual property concerns around healthcare data usage. Addressing these privacy challenges, a novel approach known as Unlearnable Examples (UEs) has been introduced, aiming to make data unlearnable to deep learning models. A prominent method within this area, called Unlearnable Clustering (UC), has shown improved UE performance with larger batch sizes but was previously limited by computational resources (e.g., a single workstation). To push the boundaries of UE performance with theoretically unlimited resources, we scaled up UC learning across various datasets using Distributed Data Parallel (DDP) training on the Summit supercomputer. Our goal was to examine UE efficacy at high-performance computing (HPC) levels to prevent unauthorized learning and enhance data security, particularly exploring the impact of batch size on UE’s unlearnability. Utilizing the robust computational capabilities of the Summit, extensive experiments were conducted on diverse datasets such as Pets, MedMNist, Flowers, and Flowers102. Our findings reveal that both overly large and overly small batch sizes can lead to performance instability and affect accuracy. However, the relationship between batch size and unlearnability varied across datasets, highlighting the necessity for tailored batch size strategies to achieve optimal data protection. The use of Summit’s high-performance GPUs, along with the efficiency of the DDP framework, facilitated rapid updates of model parameters and consistent training across nodes. Our results underscore the critical role of selecting appropriate batch sizes based on the specific characteristics of each dataset to prevent learning and ensure data security in deep learning applications. The source code is publicly available at https: // github. com/ hrlblab/ UE_ HPC .

Zhu, Yanfan [Vanderbilt University, Nashville, TN,↗

Generation and representation of synthetic smart meter data

Advanced energy algorithms running at big-data scale will be necessary to identify, realize, and verify energy savings to meet government and utility goals of building energy efficiency. Any algorithm must be well characterized and validated before it is trusted to run at these scales. Smart meter data from real buildings will ultimately be required for the development, testing, and validation of these energy algorithms and processes. However, for initial development and testing, smart meter data are difficult to work with due to privacy restrictions, noise from unknown sources, data accessibility, and other concerns which can complicate algorithm development and validation. This paper describes a new methodology to generate synthetic smart meter data of electricity use in buildings using detailed building energy modeling, which aims to capture the variability and stochastics of real energy use in buildings. The methodology can create datasets tailored to represent specific scenarios with known truth and controllable amounts of synthetic noise. Knowledge of ground truth also allows the development and validation of enhanced processes which leverage building metadata, such as building type or size (floor area), in addition to smart meter data. The methodology described in this paper includes the key influencing factors of real-world building energy use including weather data, occupant-driven loads, building operation and maintenance practices, and special events. Data formats to support workflows leveraging both synthetic meter data and associated metadata are proposed and discussed. Finally, example use cases of the synthetic meter data are described to illustrate potential applications.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Cybersecurity impact of the growth of data in sports

As the sporting industry continues to increase dependence on new and often massive data sets, they also risk exposing themselves, as well as the leagues, teams, venues, players, and fans, to risks posed by not protecting the confidentiality and integrity of this data. The industry, especially true for professional sports, demands an ever increasing supply of new forms of data to achieve their goals of enticing new fans; enhancing experience; expanding the fan experience from venues and television to the internet and mobile devices; measuring and evaluating the performance and safety of players; and increasing revenue. New methods have emerged to collect data about a variety of aspects of sports ranging from players to fans to teams to venues. Data collected during sporting events is combined with historical data to also help teams analyze and predict performance. To synthesize these predictive analytical and historical memories, sports teams and organizations mix a variety of computer technologies. System integration tends to focus on ensuring the availability of the data and reliability of the technologies but might introduce risks to the confidentiality (privacy) and integrity of the data. Recognizing the potential risks requires review and evaluation. Much of the data is collected for publicly providing information to the fans, so threats to confidentiality pose fewer risks. The greatest risk is violating the privacy of players, such a private performance test, or using technology to gain a competitive advantage.

Jenkins, Stephanie↗