SEARCH · Engineering Papers

Results for “latency”

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 415 records · Page 23

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Real-Time Artificial Intelligence for Particle Reconstruction and Higgs Physics

With the discovery of the Higgs boson at the CERN LHC, the world's highest-energy particle accelerator complex, scientists have acquired an important tool to study the fundamental building blocks of the universe. Precision measurements of Higgs bosons produced with large momentum allow for unique insights into the structure of the interactions of the Higgs boson with other particles that may shed light on physics beyond the standard model. While experimentally challenging, exploring such interactions with novel artificial intelligence (AI) methods can advance our understanding of the Higgs sector, including the Higgs boson's self-interaction. Moreover, the LHC is undergoing a major upgrade to further increase its particle collision rate and thereby operate for an additional decade. The experimental detectors at the upgraded facility must process at least a factor of ten more data at rates of hundreds of terabytes per second all under challenging conditions. New AI techniques are required to reconstruct and select, or trigger on, the most physics-sensitive events in real-time to handle the resulting avalanche of data. The proposed research will achieve the goals of the LHC program at the CMS experiment by developing a sub-microsecond event reconstruction system using real-time AI algorithms that employ field-programmable gate array technologies. By harnessing sophisticated AI techniques, this research focuses on measuring the production of Higgs bosons at large momentum while enhancing particle reconstruction methods in the trigger and beyond. Overall, the proposed research has broader implications for the use of AI in resource-constrained, low-latency embedded applications across all fields of science.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2025

Considerations for Distributed Edge Data Centers and Use of Building Loads to Support Large Interconnections

The rapid expansion of artificial intelligence (AI) and machine learning is driving unprecedented electricity demand from data centers. It is predicted that by 2030, 90% of AI workloads will be inference-based, requiring interconnection of multiple low-latency edge data centers (<20 MW) sited closer to end users - often on already constrained distribution feeders. Although individually small, these loads can aggregate to large loads per feeder, straining infrastructure, creating multi-year interconnection delays, and driving up customer costs. This paper proposes a data center-focused grid-integration framework that combines feeder hosting capacity analysis with building energy efficiency, building load flexibility, and waste heat reuse to expand effective feeder and substation headroom. Such approaches can reduce interconnection delays, lower costs for ratepayers, and accelerate AI-ready infrastructure deployment.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

DOE OSTI2025

Fast Adaptive Neural Control of Resonant Extraction at Fermilab

We present the development of a machine learning (ML) based regulation system for third-order resonant beam extraction in the Mu2e experiment at Fermilab. Classical and ML-based controllers have been optimized using semi-analytic simulations and evaluated in terms of regulation performance and training efficiency. We compare several controller architectures and discuss the integration of neural control into an adaptive framework. We also present progress on surrogate models that predict the controller response given a spill intensity and controller action history. To enable real-time deployment, we report progress on implementing low-latency, edge-based inference suitable for hardware-constrained environments. Our results demonstrate the feasibility and advantages of ML-based control in managing complex, time-varying physical systems, with broader implications for accelerator operations and other domains requiring fast, adaptive regulation.

Berlioz, Jose Rene [Fermilab]↗

DOE OSTI2026

Bridging Cloud and Edge Computing at NREL Using CONNECT: Cloud Optimized Networking for Next-Gen Edge Computing Technologies [Slides]

CONNECT is an innovative on-premise hardware and software solution that integrates edge and cloud computing infrastructure at NREL. Built on the AWS Greengrass middleware and leveraging the MQTT protocol, CONNECT enables real-time data streaming from IoT devices and gateways to both cloud and local services, empowering researchers to rapidly capture, analyze, and act upon edge-generated data while leveraging cloud capabilities. The platform addresses research infrastructure challenges by providing a pre-approved platform which is already configured with the correct networking and cybersecurity baselines thus eliminating procurement delays and enabling on-demand availability. CONNECT's hybrid architecture efficiently manages burstable workloads, allowing research teams to dynamically scale computational capacity, handle peak data loads, and reduce operational bottlenecks. Advanced capabilities include built-in GPU support for executing machine learning models which enables low-latency inference at the edge from models trained in the cloud. This architecture supports real-time analytics and filtering, providing a mechanism to allow only transmitting and processing high-value data. Cloud-based configuration management permits engineers to manage on-premise systems remotely, optimizing operational efficiency. By bridging edge and cloud computing, CONNECT provides NREL researchers with a flexible, scalable platform that accelerates scientific discovery while maintaining robust security and performance standards.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Grid Operator Analytics and Assessment Tools for Inverter- Based Resources Dominated Grid (GOAAT-IBR) Project Update

This presentation provides an update on the OPTIMA GOAAT project, with emphasis on the cloud-native data platform developed in-house to ingest, manage, and operationalize high-resolution power system data. Since our last NASPI presentation, accessible via OSTI ID #2671437, the project team advanced the design and deployment of a scalable architecture capable of handling both synchronized and non-synchronized streams, including PMU, point-on-wave (POW), COMTRADE, and SCADA data. These materials review the project status, recent progress, and key lessons learned. The core of the presentation examines the architecture and engineering of our cloud-native ingestion and data management platform. We then explain how pipelines were designed to collect, normalize, time-align, store, and serve heterogeneous data at scale. We will discuss design choices such as data models, streaming versus batch ingestion, storage tiers, and interoperability with analytics applications. Practical experiences with cloud-native technologies were shared during the event, including benefits, limitations, and integration challenges in a utility environment, along with methods used to improve performance, reduce latency, and optimize resource usage. The presentation also showcases user interface designs and visualization tools that convert raw measurements and analytics results into intuitive, actionable insights for operators and engineers. During the presentation examples were provided demonstrating how visualization, event views, and summarized analytics enhance situational awareness and support operational decision-making. These use cases illustrate how a well-designed data infrastructure can bridge the gap between high-volume measurements and practical grid operations.

Aminifar, Farrokh↗

DOE OSTI2025

LDMS New Features for Deployment in Advanced Environments and Feedback for Operations

We describe how LDMS is being used to collect application data concurrent with system data and how the low-latency availability of this data for analysis can be used for real-time data analysis and feedback in order to support efficient, resilient, and reliable system operations. Finally, we will describe current related research areas.

Brandt, James Michael [Sandia National Laboratorie↗

DOE OSTI2026

SuperLab 2.0 Showcase: Connecting Five Labs to Tackle Grid Complexity and Unlock Unique Grid Asset Potential

SuperLab 2.0 (a five-lab demonstration) is a collaborative, national-scale experiment showcasing the coordination of geographically distributed energy assets in real time. The demonstration integrates 25 physical and digital assets, spanning wind, photovoltaics, batteries, electrolyzers, DC fast chargers, microgrid controllers, building automation systems, small modular reactors, control centers, and gas turbines, across five U.S. Department of Energy (DOE) national laboratories: National Laboratory of the Rockies (NLR), Idaho National Laboratory, National Energy Technology Laboratory, Lawrence Berkeley National Laboratory, and Sandia National Laboratories. These assets are unified using Energy Sciences Network (ESnet), a low-latency, high-performance DOE network, and are controlled via a centralized energy controller hosted at NLR's Advanced Research on Integrated Energy Systems facility. The demonstration validates the ability to stress-test hybrid energy systems under dynamic scenarios to de-risk advanced control strategies for greater resilience and flexibility.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2025

Best Practices for Grid Communications

As the grid evolves, the communications architecture will need to evolve with it. That architecture affords a structured means by which the evolving complexities of the modern electric grid can be managed. This document provides best practices that can be implemented in the grid of today and evolve towards the grid and grid architecture of the future. The evolving grid and its control communications increasingly rely on commercial communications providers and a variety of technologies, from wireless (e.g., 5G, microwave, Wi-Fi) to wireline (fiber, copper) to radio communications (P25, other repeater-based systems), and all these communications systems rely on electric power. A reliable and resilient grid must account for this complex set of interdependencies in its planning activities, especially those involving restoration and recovery. The participation of all relevant parties in both planning and exercising of plans can prevent unexpected conditions that impede the reliable operation and recovery of the grid. Best practices for grid communications include using a Network Management System to document the operational state, define and monitor baselines, detect changes, and accelerate response to abnormalities. If transitioning from SONET to IP/packet-based systems, translating grid requirements into communications requirements for latency, bandwidth and throughput, IP packet delay variation, packet loss, and availability should inform and drive technology planning and selection as well as that communication system’s Quality of Service (QoS) policies and Service Level Agreements (SLAs). Secure and reliable timing is another key component of a reliable and resilient grid that can operate through adverse events. A trusted internal NTP configuration, an integrated and diverse timing delivery system, optimizing the timing architecture based on the transport technologies of the communications system, and using established standards can deliver the level of timing accuracy required by a range of time-sensitive power system applications.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2026

Intelligent Experiments Through Real-time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future EIC Detectors (Final Report)

The overall vision of this project was to integrate real-time artificial intelligence (AI) directly into the data acquisition and detector-control systems of nuclear physics experiments, including both fast online event selection and an autonomous detector-control feedback loop. The work carried out under the award focused on the fast online event-selection half of that vision: the efficient recording of low-momentum heavy-flavor (HF) hadron decays in proton-proton collisions at the sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC)—an observable that requires fast tracking and topological trigger selection not previously demonstrated at RHIC, and that is essential for QCD studies at future facilities such as the Electron-Ion Collider (EIC). The autonomous detector-control (GPU-based feedback) component named in the project title remained a design concept and was not implemented under this award. The Massachusetts Institute of Technology (MIT) group led the offline simulation and data processing needed to train the machine-learning (ML) models, the translation of trained models to Field-Programmable Gate Array (FPGA) firmware using the hls4ml framework, and the physics validation of heavy-flavor reconstruction. Over the award period, the team developed and hardware-tested the principal components of an AI-based heavy-flavor trigger on simulated and recorded sPHENIX tracker data: a software Bipartite Graph Attention Network (BiGAT) trigger model reaching > 95% signal efficiency at 99% background rejection; an FPGA-native hit clusterizer matching the offline clustering; smaller networks synthesized to FPGA within the required sub-10 µs latency; and an assembled decoder–clusterizer–inference firmware chain exercised on the FELIX readout board. A complete, fully integrated hardware demonstrator was not finished within the award period. This report documents the project goals, the MIT group’s contributions, the technical accomplishments, and the outlook toward applications at the future EIC ePIC detector.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2026

CMS Outer Tracker Module QA

The High Luminosity Large Hadron Collider is a planned upgrade to the LHC that will allow it to take up to 7 times as much data. Upgrades to all detector systems, including the Outer Tracker, are necessary to allow them to handle the increased radiation and pileup. Fermilab has been tasked with assembling modules for the Outer Tracker system. This presentation discusses the QA tests performed on these modules and the integration procedure required to ensure all modules work in sync. The QA tests include sensor IV curves, MaPSA testing, visual inspection, skeleton testing, and module testing. There are additional tests that can only be performed on integrated modules that are working in sync, and developing an understanding of the timing and latency of the modules is necessary to perform this synchronization.

Hobbeheydar, Samuel [Fermilab; Carnegie Mellon U. ↗

DOE OSTI2026

CMS Outer Tracker Module QA

Hobbeheydar, Samuel [Fermilab; Carnegie Mellon U. ↗

DOE OSTI2026

Cloud-Based Demonstration of the Eastern Interconnection Situational Awareness Monitoring System (ESAMS)

This report describes a cloud-based implementation and field demonstration of the Eastern Interconnection Situational Awareness and Monitoring System (ESAMS). ESAMS was developed to support the detection and source localization of forced oscillations using synchrophasor measurements from tie-lines connecting areas served by different reliability coordinators (RCs), so that RCs could better coordinate their response to wide-area events. A previous effort had identified deployment barriers associated with hosting shared situational awareness tools at a single RC. To address these barriers, ESAMS was migrated to Amazon Web Services and evaluated in a six-month field demonstration. ISO New England (ISO-NE) and PJM streamed data to the platform using AWS Direct Connect and a site-to-site VPN, respectively. The resulting multi-utility measurement footprint enabled regional source localization across major portions of the U.S. Eastern Interconnection and supported routine identification of oscillation events. During the final three months of the trial, 24 events above 2 MW/MVAR were detected. The largest detected oscillation approached a 25 MW peak-to-peak amplitude, and the longest persisted intermittently for more than 11 hours. The demonstration also assessed operational considerations—including data transfer volumes, end-to-end latency, and cloud computing costs—and found that network and compute requirements were modest relative to typical cloud capabilities while providing performance comparable to prior on-premises deployments. Overall, the results indicate that cloud hosting can provide a practical path to shared interconnection-wide oscillation monitoring. The cloud ESAMS demonstration establishes a foundation for broader utility participation and for building future wide-area analytics that leverage measurements across organizational boundaries.

Follum, James D.↗

DOE OSTI2026

A Modular Accelerator Robotics Framework for AD Robotics

Accelerator tunnels, such as the ones at Fermilab, remain highly radioactive after beam shutoff due to induced radiation from the beam. This residual radiation creates a hazardous environment for manual inspection and repair of beamline components. To minimize worker radiation dose and reduce beam downtime, the AD Robotics Initiative previously built a fleet of low-cost custom mobile robots. However, the custom Python sockets server-client architecture lacked standardization, causing development delays and complicating the integration of new sensors and actuators. Here, we developed a modular system using ROS2 and Docker to standardize the teleoperation and control interfaces. This system was validated by implementing a teleoperation controller with real-time, low-latency, and high-definition video feedback. The aim of this framework is for a new feature or even a robot to be integrated into the system simply by documenting the hardware configuration. Current integration of LiDAR, Odometry, and Depth Cameras provides the foundation for Simultaneous Localization and Mapping (SLAM) tasks. Finally, future work involves integration into the accelerator control system and the attachment of a 6 degree-of-freedom robotic arm for telemanipulation.

Rayyan Khan, M. [Fermilab; Rensselaer Poly.; Unlis↗

DOE OSTI2022

Real-time GIS Programming and Geocomputation

Streaming data generated continuously from sensor networks, mobile devices, social media platforms and other edge devices have posed significant challenges to existing computing platforms for achieving both high throughput and low latency data processing in addition to scalable computing. This entry introduces a real-time computing and programming platform for time-critical GIS (Geographic Information System) applications. In this platform, advanced streaming data processing software, such as Apache Kafka and Spark Streaming, are integrated to enable data analytics in real-time. This computing platform can also be extended to integrate GeoAI (Geospatial Artificial Intelligence) based machine learning models to leverage both historical and streaming data to achieve real-time prediction and intelligent geospatial analytics. Two real-time geospatial applications in terms of flood simulation and climate data visualization are introduced to demonstrate how real-time programming and computing can help tackle real-world problems with important societal impacts.

Liu, Yan↗

DOE OSTI2022

A Logarithmic Bayesian Approach to Quantum Error Detection

We consider the problem of continuous quantum error correction from a Bayesian perspective, proposing a pair of digital filters using logarithmic probabilities that are able to achieve near-optimal performance on a three-qubit bit-flip code, while still being reasonable to implement on low-latency hardware. These practical filters are approximations of an optimal filter that we derive explicitly for finite time steps, in contrast with previous work that has relied on stochastic differential equations such as the Wonham filter. By utilizing logarithmic probabilities, we are able to eliminate the need for explicit normalization and can reduce the Gaussian noise distribution to a simple quadratic expression. The state transitions induced by the bit-flip errors are modeled using a Markov chain, which for log-probabilities must be evaluated using a LogSumExp function. We develop the two versions of our filter by constraining this LogSumExp to have either one or two inputs, which favors either simplicity or accuracy, respectively. Using simulated data, we demonstrate that the single-term and two-term filters are able to significantly outperform both a double threshold scheme and a linearized version of the Wonham filter in tests of error detection under a wide variety of error rates and time steps.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

ExtremeMETA: High-speed Lightweight Image Segmentation Model by Remodeling Multi-channel Metamaterial Imagers

Deep neural networks (DNNs) have heavily relied on traditional computational units, such as CPUs and GPUs. However, this conventional approach brings significant computational burden, latency issues, and high power consumption, limiting their effectiveness. This has sparked the need for lightweight networks such as ExtremeC3Net. Meanwhile, there have been notable advancements in optical computational units, particularly with metamaterials, offering the exciting prospect of energy-efficient neural networks operating at the speed of light. Yet, the digital design of metamaterial neural networks (MNNs) faces precision, noise, and bandwidth challenges, limiting their application to intuitive tasks and low-resolution images. In this study, we proposed a large kernel lightweight segmentation model, ExtremeMETA. Based on ExtremeC3Net, our proposed model, ExtremeMETA maximized the ability of the first convolution layer by exploring a larger convolution kernel and multiple processing paths. With the large kernel convolution model, we extended the optic neural network application boundary to the segmentation task. To further lighten the computation burden of the digital processing part, a set of model compression methods was applied to improve model efficiency in the inference stage. The experimental results on three publicly available datasets demonstrated that the optimized efficient design improved segmentation performance from 92.45 to 95.97 on mIoU while reducing computational FLOPs from 461.07 MMacs to 166.03 MMacs. The large kernel lightweight model ExtremeMETA showcased the hybrid design’s ability on complex tasks.

large convolution kernel↗

DOE OSTI2023

FPGA Acceleration of GCN in Light of the Symmetry of Graph Adjacency Matrix

Graph Convolutional Neural Networks (GCNs) are widely used to process large-scale graph data. Different from deep neural networks (DNNs), GCNs are sparse, irregular, and unstructured, posing unique challenges to hardware acceleration with regular processing elements (PEs). In particular, the adjacency matrix of a GCN is extremely sparse, leading to frequent but irregular memory access, low spatial/temporal data locality and poor data reuse. Furthermore, a realistic graph usually consists of unstructured data (e.g., unbalanced distributions), creating significantly different processing times and imbalanced workload for each node in GCN acceleration. To overcome these challenges, we propose an end-to-end hardware-software co-design to accelerate GCNs on resource-constrained FPGAs with the features including: (1) A custom dataflow that leverages symmetry along the diagonal of the adjacency matrix to accelerate feature aggregation for undirected graphs. We utilize either the upper or the lower triangular matrix of the adjacency matrix to perform aggregation in GCN to improve data reuse. (2) Unified compute cores for both aggregation and transform phases, with full support to the symmetry-based dataflow. These cores can be dynamically reconfigured to the systolic mode for transformation or as individual accumulators for aggregation in GCN processing. (3) Preprocessing of the graph in software to rearrange the edges and features to match the custom dataflow. This step improves the regularity in memory access and data reuse in the aggregation phase. Moreover, we quantize the GCN precision from FP32 to INT8 to reduce the memory footprint without losing the inference accuracy. We implement our accelerator design in Intel Stratix10 MX FPGA board with HBM2, and demonstrate 1.3x-110.5x improvement in end-to-end GCN latency as compared to the state-of the-art FPGA implementations, on the graph datasets of Cora, Pubmed, Citeseer and Reddit.

Nair, Gopikrishnan R.↗

DOE OSTI2024

Optimizing Distributed Training on Frontier for Large Language Models

Large language models (LLMs) have demonstrated remarkable success as foundational models, benefiting various downstream applications through fine-tuning. Loss scaling studies have demonstrated the superior performance of larger LLMs compared to their smaller counterparts. Nevertheless, training LLMs with billions of parameters poses significant challenges and requires considerable computational resources. For example, training a one trillion parameter GPT-style model on 20 trillion tokens requires a staggering 120 million exaflops. This research explores efficient distributed training strategies to extract this computation from Frontier, the world's first exascale supercomputer. We enable and investigate various model and data parallel training techniques, such as tensor parallelism, pipeline parallelism, and sharded data parallelism, to facilitate training a trillion-parameter model on Frontier. We empirically assess these techniques and their associated parameters to determine their impact on memory footprint, communication latency, and GPU's computational efficiency. We analyze the complex interplay among these techniques and find a strategy to combine them to achieve high throughput through hyperparameter tuning. We have identified efficient strategies for training large LLMs of varying sizes through empirical analysis and hyperparameter tuning. For 22 Billion, 175 Billion, and 1 Trillion parameters, we achieved GPU throughputs of 38.38%, 36.14%, and 31.96%, respectively. For the training of the 175 Billion parameter model and the 1 Trillion parameter model, we achieved 100% weak scaling efficiency on 1024 and 3072 Mi250X GPUs, respectively. We also achieved strong scaling efficiencies of 89% and 87% for these two models. We trained these models only tens of iterations instead of training till completion.

Yin, Junqi↗