SEARCH · Search NASA

Results for “replay”

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.

SourceResearch connectionFrom yearThrough year

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

At least 55 records · Page 3

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Performance Evaluation of Vertical Federated Machine Learning Against Adversarial Threats on Wide-Area Control System: Preprint

Federated machine learning (FL) is gaining significant popularity to develop cybersecurity solutions in power grids because of its advanced capability to support decentralized data handing at local devices, its privacy preservation, and its low-bandwidth requirement. However, the evolving adversarial machine learning (AML) threats raise significant concerns for the cybersecurity of FL architectures. The FL-based split neural network (SplitNN) achieves high performance through the decentralized training of local neural network models while preserving data privacy across multiple entities. In this paper, we propose a methodology for evaluating the performance of a vertical FLbased anomaly detector against different types of AML attacks, including denial-of-service attacks, adversarial data injection attacks, and replay attacks on the trained local models deployed in the grid network. For a case study, we consider the modified IEEE 13-bus system, and we develop SplitNN-based binary and multiclass classification models to detect, locate, and identify different types of data integrity attacks on the volt-watt control with two pooling layers: maximum pooling and AvgPool. Our experimental results, computed through performance metrics, reveal that the severity of these AML attacks varies with the integrated pooling mechanism, the type of classification model, and the nature of the cyberattack. Further, the AML attacks negatively impacted the prediction time per sample for the pretrained SplitNN during the online testing.

adversarial threats↗

DOE OSTI2024

Mneme

A simple tool allowing recording the execution of a GPU (CUDA) kernel and replaying that kernel as an independent executable. The tool operates in 3 phases. During compile time the user needs to apply a provided LLVM pass to instrument the code. The pass detects all device global variables and device functions and stores this information with the respective LLVM-IR in the global device memory. The compilation generates a record-able executable. The second phase involves running the application executable with a desired input and using LD_PRELOAD to enable recording. When recording before invoking a device kernel the pre-loaded library stores device memory in persistent storage and associates the memory with the device kernel and an LLVM IR file. At the end of the recorded execution the pre-load library generates a database in the form of a JSON file containing information regarding the LLVM-IR files and the snapshots of device memory. During the third and last phase the user can replay the execution of an kernel as a separate independent executable. Besides executing it the user can modify the LLVM IR file and auto-tune parameters such as kernel launch-bounds or kernel runtime execution parameters (e.g. Kernel Block and Grid Dimensions). Is

Parasyris, Konstantinos↗

DOE OSTI2023

Hashes are not suitable to verify fixity of the public archived web

Web archives, such as the Internet Archive, preserve the web and allow access to prior states of web pages. We implicitly trust their versions of archived pages, but as their role moves from preserving curios of the past to facilitating present day adjudication, we are concerned with verifying the fixity of archived web pages, or mementos, to ensure they have always remained unaltered. A widely used technique in digital preservation to verify the fixity of an archived resource is to periodically compute a cryptographic hash value on a resource and then compare it with a previous hash value. If the hash values generated on the same resource are identical, then the fixity of the resource is verified. We tested this process by conducting a study on 16,627 mementos from 17 public web archives. We replayed and downloaded the mementos 39 times using a headless browser over a period of 442 days and generated a hash for each memento after each download, resulting in 39 hashes per memento. The hash is calculated by including not only the content of the base HTML of a memento but also all embedded resources, such as images and style sheets. We expected to always observe the same hash for a memento regardless of the number of downloads. However, our results indicate that 88.45% of mementos produce more than one unique hash value, and about 16% (or one in six) of those mementos always produce different hash values. We identify and quantify the types of changes that cause the same memento to produce different hashes. These results point to the need for defining an archive-aware hashing function, as conventional hashing functions are not suitable for replayed archived web pages.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

DOE OSTI2021

Optimizing and Extending the Functionality of EXARL for Scalable Reinforcement Learning [Slides]

The main goal of the Co-Design Summer School 2021 is to provide algorithmic improvements to EXARL framework by improving performance and by adding functionalities. This presentation includes an introduction to reinforcement learning and to EXARL. The researchers expanded the capability of EXARL by including additional agents like (Asynchronized) Advantage Actor Critic (A2C/A3C) and Twin Delayed Deep Deterministic Policy Gradient (TD3). They also explored algorithmic improvements such as v-trace and Prioritized Experience Replay. They found that A2C/A3C performed best with v-trace and outperformed Deep Q-Network (DQN) on both the CartPole game and the ExaBooster scientific environment. Additionally, they found that TD3 performed as good as the existing Deep Deterministic Policy Gradient (DDPG) agent and that adding Prioritized Experience Replay to DDPG accelerated convergence.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

Edge ML for CAN bus intrusion detection in AVs

Autonomous Vehicles (AVs) are revolutionizing transportation, but their reliance on interconnected cyber-physical systems exposes them to unprecedented cybersecurity risks. This study addresses the critical challenge of detecting real-time cyber intrusions in self-driving vehicles by leveraging a dataset from the Udacity self-driving car project. We simulate four high-impact attack vectors, Denial of Service (DoS), spoofing, replay, and fuzzy attacks, by injecting noise into spatial features (e.g., bounding box coordinates) to replicate adversarial scenarios. We develop and evaluate two lightweight neural network architectures (NN-1 and NN-2) alongside a logistic regression baseline (LG-1) for intrusion detection. The models achieve exceptional performance, with NN-2 attaining an AUC score of 93.15% and 93.15% accuracy, demonstrating their suitability for edge deployment in AV environments. Through explainable AI techniques, we uncover unique forensic fingerprints of each attack type, such as spatial corruption in fuzzy attacks and temporal anomalies in replay attacks, offering actionable insights for feature engineering and proactive defense. Visual analytics, including confusion matrices, ROC curves, and feature importance plots, validate the models' robustness and interpretability. This research sets a new benchmark for AV cybersecurity, delivering a scalable, field-ready toolkit for Original Equipment Manufacturers (OEMs) and policymakers. By aligning intrusion fingerprints with SAE J3061 automotive security standards, we provide a pathway for integrating machine learning into safety-critical AV systems. Our findings underscore the urgent need for security-by-design AI, ensuring that AVs not only drive autonomously but also defend autonomously. This work bridges the gap between theoretical cybersecurity and life-preserving engineering, offering a leap toward safer, more secure autonomous transportation.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS1994

Delta/Wind Launch with Isolated Cameras from Continuous Recording

The Wind spacecraft represents one of NASA's contributions to the International Solar Terrestrial Program (ISTP), an international effort to quantify the effects of solar energy on the Earth's magnetic field. Wind will provide continuous measurement of the solar wind, particularly charged particles and magnetic field data. The specific objectives of Wind are to: (1) provide complete plasma, energetic particle, and magnetic field input for magnetospheric and ionospheric studies; (2) determine the magnetospheric output to interplanetary space in the upstream region; (3) investigate basic plasma processes occurring in the near-Earth solar wind; and (4) provide baseline ecliptic plane observations to be used in heliospheric studies. This videotape shows the pre-dawn launch of the Wind spacecraft aboard a Delta 7925 on November 1, 1994. After the countdown and launch, the tape shows the activity in the Telemetry Room at Kennedy Space Center, where people are following the progress of the spacecraft. Following the activity in the telemetry room, there are four different replays of the launch from different locations. After showing the replays of the launch, the video returns to the Telemetry Room when an important stage in the launch and flight is achieved.

Source record↗

NASA NTRS2002

STS-111 Flight Day 3 Highlights

On Flight Day 3 of STS-111, the crew of Endeavour (Kenneth Cockrell, Commander; Paul Lockhart, Pilot; Franklin Chang-Diaz, Mission Specialist; Philippe Perrin, Mission Specialist) and the Expedition 5 crew (Valery Korzun, Commander; Peggy Whitson, Flight Engineer; Sergei Treschev, Flight Engineer), begin their final approach towards the International Space Station (ISS). From cameras aboard the ISS, live video of Endeavour is shown as it approaches the station. The Orbiter is maneuvered slowly to a position for docking, and cameras from multiple angles show this process. As it is maneuvered, there are clear views of its payload bay, which includes the Leonardo MultiPurpose Logistics Module (MPLM) and the Mobile Base System (MBS), both of which will be installed on the ISS during this mission. In the final stages of the docking procedure there is close-up footage of Endeavour meeting the ISS's Pressurized Mating Adapter 2 on the Destiny Laboratory Module. Inside the ISS, the Expedition 4 crew (Yury Onufrienko, Commander; Daniel Bursch, Flight Engineer; Carl Walz, Flight Engineer), which will be replaced by the Expedition 5 crew, prepares for final docking. Crew members are shown transferring equipment from the Endeavour to the ISS, prior to a replay of the mating of the two crafts. In the replay, the hatch is shown being opened and the two newly arrived crews are greeted with excitement by Expedition 4 crewmembers. The video closes with footage of the Quest airlock used for EVA (extravehicular activity) egress, and the Canadarm 2 robotic arm.

Source record↗

NASA NTRS2002

STS-110/Atlantic/ISS 8A Pre-Launch On Orbit-Landing-Crew Egress

The crew of STS-110, which consists of Commander Michael Bloomfield, Pilot Stephen Frick, and Mission Specialists Rex Walheim, Ellen Ochoa, Lee Morin, Jerry Ross, and Steven Smith is introduced at the customary pre-flight meal. The narrator provides background information on the astronauts during suit-up. Each crew member is shown in the White Room before boarding Space Shuttle Atlantis, and some display signs to loved ones. Launch footage includes the following replays: Beach Tracker, VAB, Pad B, Tower 1, DLTR-3, Grandstand, Cocoa Beach DOAMS, Playalinda DOAMS, UCS-23, SLF Convoy, OTV-154, OTV-163, OTV-170 (mislabeled), and OTV-171 (mislabeled). After the launch, NASA administrator Sean O'Keefe gives a speech to the Launch Control Center, with political dignitaries present. While on-orbit, Atlantis docks with the International Space Station (ISS), and Canadarm 2 on the ISS lifts the S0 Truss out of the orbiter's payload bay. The video includes highlights of three extravehicular activities (EVAs). In the first, the S0 Truss is fastened to the Destiny Laboratory Module on the ISS. During the third EVA, Walheim and Smith assist in the checkout of the handcart on the S0 Truss. The Atlantis crew is shown gathered together with the Expedition 4 crew of the ISS, and again by itself after undocking. Replays of the landing include: VAB, Tower 1, Mid-field, Runway South End, Runway North End, Tower 2, Playalinda DOAMS, Cocoa Beach DOAMS, and Pilot Point of View (PPOV). After landing, Commander Bloomfield lets each of his crew members give a short speech.

Source record↗

NASA NTRS2001

STS-105/Discovery/ISS 7A.1: Pre-Launch Activities, Launch, Orbit Activities and Landing

The crew of Space Shuttle Discovery on STS-105 is introduced at their pre-launch meal and at suit-up. The crew members include Commander Scott Horowitz, Pilot Rick Sturckow, and Mission Specialists Patrick Forrester and Daniel Barry, together with the Expedition 3 crew of the International Space Station (ISS). The Expedition 3 crew includes Commander Frank Culbertson, Soyuz Commander Vladimir Dezhurov, and Flight Engineer Mikhail Tyurin. When the astronauts depart for the launch pad in the Astrovan, their convoy is shown from above. Upon reaching the launch pad, they conduct a walk around of the shuttle, display signs for family members while being inspected in the White Room, and are strapped into their seats onboard Disciovery. The video includes footage of Discovery in the Orbiter Processing Facility, and some of the pre-launch procedures at the Launch Control Center are shown. The angles of launch replays include: TV-1, Beach Tracker, VAB, Pad A, Tower 1, UCS-15, Grandstand, OTV-70, Onboard, IGOR, and UCS-23. The moment of docking between Discovery and the ISS is shown from inside Discovery's cabin. While in orbit, the crew conducted extravehicular activities (EVAs) to attach an experiments container, and install handrails on the Destiny module of the ISS. The video shows the docking and unloading of the Leonardo Multipurpose Logistics Module (MPLM) onto the ISS. The deployment of a satellite from Discovery with the coast of the Gulf of Mexico in the background is shown. Cape Canaveral is also shown from space. Landing replays include VAB, Tower 1, mid-field, South End SLF, North End SLF, Tower 2, Playalinda DOAMS, UCS-23, and Pilot Point of View (PPOV). NASA Administrator Dan Goldin meets the crew upon landing and participates in their walk around of Discovery. The video concludes with a short speech by commander Horowitz.

Source record↗

NASA NTRS2019

Quantifying Contributions of Land Carbon Fluxes Variability and Atmospheric Transport Variability to Atmospheric CO2 Variability

Much of the uncertainty in predicting variations in greenhouse gases originates in the complex dynamics of the land ecosystem and the atmosphere. To reduce the uncertainties, it is useful to decouple the contributions of land carbon fluxes and atmospheric transport to atmospheric carbon variability. Here we isolate these contributions using a version of the NASA GEOS model that couples carbon, energy and water cycles between the land and the atmosphere. Our current study is a follow-on to a preliminary analysis that suggested that an extreme event (e.g., imposed drought) in a free running AGCM simulation affects atmospheric CO2 more through its impact on atmospheric transport than through a modification of land carbon fluxes. In order to more carefully isolate the effects of the land carbon variability and atmospheric transport variability on atmospheric CO2 variability, we conducted two coupled-AGCM simulations in replay mode, a mode that forces the model's evolution of weather to match that of the MERRA-2 reanalysis. In the control simulation, the land carbon fluxes and the atmospheric CO2 concentrations, as well as the meteorology, are simulated over 2001-2015. We compute the climatological seasonal cycles of net land carbon production from this control simulation and then prescribe, in a second simulation, these climatological carbon fluxes to the atmosphere in the same replay mode. By comparing the atmospheric CO2 variability produced in the two simulations, we fully isolate the part of this variability associated with land surface fluxes. Relative contributions of land flux variability and atmospheric transport variability to CO2 variability are quantified both on a regional basis and as a function of height into the atmosphere to support interpretation of both ground-based and satellite observations.

Lee, Eunjee↗

NASA NTRS2018

Evaluation of MERRA-2-Based Ozone Profile Simulations with the Global Ozonesonde Network

Chemical transport model (CTM) hindcasts of ozone (O3) are useful for filling in observational gaps and providing context for observed O3 variability and trends. We use global networks of ozonesonde stations to evaluate the O3 profiles in two simulations running versions of the NASA Global Modeling Initiative (GMI) chemical mechanism. Both simulations are tied to the NASA Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2) meteorological reanalysis: 1) The GMI CTM, and 2) The MERRA-2 GMI Replay (M2 GMI). Both simulations start in 1980, and are compared against >50,000 ozonesonde profiles from 37 global stations from the tropics to the poles. The comparisons allow us to evaluate how the Replay technique affects modeled O3 distribution, how an updated chemical mechanism in the GMI CTM affects simulated tropospheric O3 amounts, and how observed O3 distributions compare to the full set of model output. In general, M2 GMI O3 is ~10% higher than in the GMI CTM, and shows global near-surface and tropical upper troposphere/lower stratosphere (UT/LS) high biases. The updated chemical mechanism in the GMI CTM reduces these high biases. Both simulations show similar negative biases in tropical free-tropospheric O3, especially during typical biomass burning seasons. The simulations are highly-correlated with ozonesonde measurements, particularly in the UT/LS (r > 0.8), showing the ability of MERRA-2 to capture tropopause height variations. Both simulations show improved correlations with ozonesonde data and smaller O3 biases in recent years. We expect to use the sonde/model comparisons to diagnose causes of disagreement and to gauge the feasibility of calculating multidecadal O3 trends from the model output.

Stauffer, Ryan↗

NASA NTRS2018

Evaluation of MERRA-2-Based Ozone Profile Simulations with the Global Ozonesonde Network

Stauffer, Ryan M.↗

DOE OSTI2022

Performance for GPU exceptions

Techniques for improving performance of accelerated processing devices (“APDs”) when exceptions occur are provided. In APDs, the very large number of parallel processing execution units, and the complexity of the hardware used to execute a large number of work-items in parallel, means that APDs typically stall when an exception occurs (unlike in central processing units (“CPUs”), which are able to execute speculatively and out-of-order). However, the techniques provided herein allow at least some execution to occur past exceptions. Execution past an exception generating instruction occurs by executing instructions that would not lead to a corruption while skipping those that would lead to a corruption. After the exception has been satisfied, execution occurs in a replay mode in which the potentially exception-generating instruction is executed and in which instructions that did not execute in the exception-wait mode are executed. A mask and counter are used to control execution in replay mode.

Gutierrez, Anthony T.↗

DOE OSTI2024

Efficient distributed continual learning for steering experiments in real-time

Deep learning has emerged as a powerful method for extracting valuable information from large volumes of data. However, when new training data arrives continuously (i.e., is not fully available from the beginning), incremental training suffers from catastrophic forgetting (i.e., new patterns are reinforced at the expense of previously acquired knowledge). Training from scratch each time new training data becomes available would result in extremely long training times and massive data accumulation. Rehearsal-based continual learning has shown promise for addressing the catastrophic forgetting challenge, but research to date has not addressed performance and scalability. To fill this gap, we propose an approach based on a distributed rehearsal buffer that efficiently complements data-parallel training on multiple GPUs to achieve high accuracy, short runtime, and scalability. It leverages a set of buffers (local to each GPU) and uses several asynchronous techniques for updating these local buffers in an embarrassingly parallel fashion, all while handling the communication overheads necessary to augment input minibatches using unbiased, global sampling. We further propose a generalization of rehearsal buffers to support both classification and generative learning tasks, as well as more advanced rehearsal strategies (notably Dark Experience Replay, leveraging knowledge distillation). We illustrate this approach with a real-life HPC streaming application from the domain of ptychographic image reconstruction. Furthermore, we run extensive experiments on up to 128 GPUs of the ThetaGPU supercomputer to compare our approach with baselines representative of training-from-scratch (the upper bound in terms of accuracy) and incremental training (the lower bound). Results show that rehearsal-based continual learning achieves a top-5 validation accuracy close to the upper bound, while simultaneously exhibiting a runtime close to the lower bound.

Asynchronous data management↗

DOE OSTI2024

Investigating the effects of cooperative transmission expansion planning on grid performance during heat waves with varying spatial scales

There is growing recognition of the advantages of interregional transmission capacity to decarbonize electricity grids. A less explored benefit is potential performance improvements during extreme weather events. This study examines the impacts of cooperative transmission expansion planning using an advanced modeling chain to simulate power grid operations of the United States Western Interconnection in 2019 and 2059 under different levels of collaboration between transmission planning regions. Two historical heat waves in 2019 with varying geographical coverage are replayed under future climate change in 2059 to assess the transmission cooperation benefits during grid stress. The results show that cooperative transmission planning yields the best outcomes in terms of reducing wholesale electricity prices and minimizing energy outages both for the whole interconnection and individual transmission planning regions. Compared to individual planning, cooperative planning reduces wholesale electricity prices by 64.3 % and interconnection-wide total costs (transmission investments + grid operations) by 34.6 % in 2059. It also helps decrease greenhouse gas emissions by increasing renewable energy utilization. However, the benefits of cooperation diminish during the widespread heat wave when all regions face extreme electricity demand due to higher space cooling needs. Despite this, cooperative transmission planning remains advantageous, particularly for California Independent System Operator with significant diurnal solar generation capacity. This study suggests that cooperation in transmission planning is crucial for reducing costs and increasing reliability both during normal periods and extreme weather events. It highlights the importance of optimizing the strategic investments to mitigate challenges posed by wider-scale extreme weather events of the future.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2022

Impact of cyber attacks on distributed compressive sensing based state estimation in power distribution grids

Modern power distribution grids suffer from multiple vulnerabilities due to the tight integration between the physical system and the cyber infrastructure. Sophisticated and malicious cyber attacks continue to adversely impact the grid operation leading to performance degradation, service interruption, and grid failure. State estimation plays an essential role in grid monitoring and advancing cyber-attack situational awareness. In this regard, this paper first proposes a distributed compressive sensing (CS) state estimation approach for an unobservable distribution grid. Further, the proposed distributed CS approach divides the distribution grid into sub-areas to perform local state estimation. Then an alternating direction method of multipliers (ADMM) based iterative information exchange among neighboring areas is employed to complete the estimation process. In this estimation process, the impact of loss of measurement data, false data injection (FDI), replay, and neighborhood cyber-attacks is analyzed. Extensive simulations are performed on the IEEE 37-bus and IEEE 123-bus standard networks to demonstrate the algorithm’s robustness to the aforementioned cyber-attacks. A quantitative analysis of computational complexity and simulation time of the distributed CS based approach is also presented.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2023

Microgrid energy scheduling under uncertain extreme weather: Adaptation from parallelized reinforcement learning agents

Microgrids are useful solutions for integrating renewable energy resources and providing seamless green electricity to minimize carbon footprint. In recent years, extreme weather events happened often worldwide and caused significant economic and societal losses. Such events bring uncertainties to the microgrid energy scheduling problems and increase the challenges of microgrid operation. Traditional optimization approaches suffer from the inaccuracy of the uncertain microgrid model and the unseen events. Existing reinforcement learning (RL) - based approaches are also hampered by the limited generalization and the increasing computational burden when stochastic formulations are required to accommodate the uncertainties. This paper proposes a new parallelized reinforcement learning (PRL) method based on the probabilistic events to handle the microgrid energy uncertainties. Specifically, several local learning agents are employed to interact with pertinent microgrid environments in a distributed manner and report outcomes to the global agent, which will optimize microgrid energy resources online during extreme events. The stochastic microgrid energy optimization problem is reformulated to include all possible scenarios with probabilities. The advantage estimate functions of learning agents are designed with a backward sweep to transfer the outcomes to the value function updating process. Two simulation studies, stochastic optimization and online testing, are performed to compare with several existing RL approaches. Results substantiate that the proposed PRL method can achieve up to 20% optimization performance improvement with 4 and 28 times less computation cost than Q-learning with experience replay and multi-agent Q-learning approaches, respectively.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2026

Evaluating lightweight unsupervised online IDS for masquerade attacks in CAN

Vehicular controller area networks (CANs) are susceptible to masquerade attacks by malicious adversaries. In masquerade attacks, adversaries silence a targeted ID and then send malicious frames with forged content at the expected timing of benign frames. As masquerade attacks could seriously harm vehicle functionality and are the stealthiest attacks to detect in CAN, recent work has devoted attention to compare frameworks for detecting masquerade attacks in CAN. However, most existing works report offline evaluations using CAN logs already collected using simulations that do not comply with the domain’s real-time constraints. Here we contribute to advance the state of the art by presenting a comparative evaluation of four different non-deep learning (DL)-based unsupervised online intrusion detection systems (IDS) for masquerade attacks in CAN. Our approach differs from existing comparative evaluations in that we analyze the effect of controlling streaming data conditions in a sliding window setting. In doing so, we use realistic masquerade attacks being replayed from the ROAD dataset. We show that although evaluated IDS are not effective at detecting every attack type, the method that relies on detecting changes in the hierarchical structure of clusters of time series produces the best results at the expense of higher computational overhead. We discuss limitations, open challenges, and how the evaluated methods can be used for practical unsupervised online CAN IDS for masquerade attacks.

Anomaly detection↗