SEARCH · Engineering Papers

Results for “time scale separation”

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 271 records · Page 15

All sources NASA NTRS DOE OSTI

DOE OSTI2026

Isolating the Vibrational Spectra of the Red Chlorophylls in Photosystem I with Multispectral Two-Dimensional Spectroscopy

Photosystem I (PSI) uses an antenna of chlorophyll (Chl) molecules to create a charge separated state with high quantum efficiency. Understanding the charge separation mechanism is currently hindered by spectral overlap between the antenna and reaction center (RC) Chls and the fact that energy transfer and electron transfer occur with similar time scales. Here, we characterize the antenna excited states by applying two-dimensional electronic (2DES) and two-dimensional electronic-vibrational (2DEV) spectroscopy to PSI complexes with closed RCs. Comparison of the 2DES and 2DEV spectra, which evolve with the same kinetics, enabled characterization of the vibrational modes of the antenna during energy equilibration between spectrally distinct Chls. Through global analysis, we learn how energy transfer between the Bulk and Red Chls presents in the 2DEV spectra and we definitively identify vibrations of the cationic components of the mixed exciton and intermolecular charge transfer states associated with the Red Chls. This work enables future studies of the initial charge separation mechanism of PSI by 2DEV spectroscopy.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

NASA NTRS1998

Standard Giant Branches in the Washington Photometric System

We have obtained CCD photometry in the Washington system C, T(sub 1) filters for some 850,000 objects associated with 10 Galactic globular clusters and 2 old open clusters. These clusters have well-known metal abundances, spanning a metallicity range of 2.5 dex from [Fe/H] approx -2.25 to +0.25 at a spacing of approx. 0.2 dex. Two independent observations were obtained for each cluster and internal checks, as well as external comparisons with existing photoelectric photometry, indicate that the final colors and magnitudes have overall uncertainties of 0.03 mag. Analogous to the method employed by Da Costa and Armandroff for V, I photometry , we then proceed to construct standard ((M(sub T),(C - T(sub 1))(sub 0)) giant branches for these clusters adopting the Lee et distance scale, using some 350 stars per globular cluster to define the giant branch. We then determine the metallicity sensitivity of the ((C - T(sub 1))(sub 0) color at a given M((sub T)(sub 1)) value. The Washington system technique is found to have three times the metallicity sensitivity of the V, I technique. At M((sub T)(sub 1)) = -2 (about a magnitude below the tip of the giant branch, roughly equivalent to M(sub I) = -3), the giant branches of 47 Tuc and M15 are separated by 1.16 magnitudes in (V - l)(sub 0) and only 0.38 magnitudes in (V - I)(sub 0). Thus, for a given photometric accuracy, metallicities can be determined three times more precisely with the Washington technique. We find a linear relationship between (C - T(sub l)(sub 0) (at M(sub T)(sub 1) = -2) and metallicity exists over the full metallicity range, with an rms of only 0.04 dex. We also derive metallicity calibrations for M(sub T)(sub 1) = -2.5 and -1.5, as well as for two other metallicity scales. The Washington technique retains almost the same metallicity sensitivity at faint magnitudes , and indeed the standard giant branches are still well separated even below the horizontal branch. The photometry is used to set upper limits in the range 0.03 - 0.09 dex for any intrinsic metallicity dispersion in the calibrating clusters. The calibrations are applicable to objects with ages approx. greater than 5 Gyr - any age effects are small or negligible for such objects. This new technique is found to have many advantages over the old two-color diagram technique for deriving metallicities from Washington photometry. In addition to only requiring 2 filters instead of 3 or 4, the new technique is generally much less sensitive to reddening and photometric errors, and the metallicity sensitivity is many times higher. The new technique is especially advantageous for metal-poor objects. The five metal-poor clusters determined by Geisler et al., using the old technique, to be much more metal-poor than previous indications, yield metallicities using the new technique which are in excellent agreement with the Zinn scale.

Geisler, Doug↗

DOE OSTI2020

Facile Size-Selective Defect Sealing in Large-Area Atomically Thin Graphene Membranes for Sub-Nanometer Scale Separations

Atomically thin graphene with a high-density of precise subnanometer pores represents the ideal membrane for ionic and molecular separations. However, a single large-nanopore can severely compromise membrane performance and differential etching between pre-existing defects/grain boundaries in graphene and pristine regions presents fundamental limitations. Here, we show for the first time that size-selective interfacial polymerization after high-density nanopore formation in graphene not only seals larger defects (>0.5 nm) and macroscopic tears but also successfully preserves the smaller subnanometer pores. Low-temperature growth followed by mild UV/ozone oxidation allows for facile and scalable formation of high-density (4–5.5 $\times$ 10 12 cm –2 ) useful subnanometer pores in the graphene lattice. We demonstrate scalable synthesis of fully functional centimeter-scale nanoporous atomically thin membranes (NATMs) with water (~0.28 nm) permeance ~23$\times$ higher than commercially available membranes and excellent rejection to salt ions (~0.66 nm, >97% rejection) as well as small organic molecules (~0.7–1.5 nm, ~100% rejection) under forward osmosis. $\times$

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

Integral and Separate Effects Test Facilities To Support Water Cooled Small Modular Reactors: A Review

This study reviews previous experimental facilities and test programs relevant to water-cooled reactor system design and analysis to meet regulatory compliances. This study aims to find the best solution for designing the required experiments, obtaining necessary test data, and verifying the developed computer code/models to support the new reactor design and development while minimizing cost and time while leveraging experiences from previous facilities to minimize. Nuclear reactor licensing requires supportive design, analysis, and experimental results to ensure the safety of the full-scale prototype reactor in regular operation, as well as during postulated accident scenarios. These reactor design analyses are generally performed using system codes and other associated simulation tools that require assessment, verification and validation using an appropriate experimental dataset. Experimental facilities used for reactor system safety analysis and system code assessments are categorized into integral effects test (IET) and separate effects test (SET) facilities. Further, these IET and SET experiments and studies use geometrically scaled systems to reproduce the prototype system behavior at a reasonable cost, albeit with some scaling-related distortions. The design challenge of these model facilities is to identify and minimize scaling distortions while reproducing the most important operational phenomena in steady-state operation and in postulated accident scenarios. Lessons learned from previous experimental facilities, models, and correlations can support the development of new multipurpose, scaled, hybrid, integrated, and modular experimental facilities for advanced light water-cooled small modular reactors (SMRs). Successful operation of these facilities can significantly reduce upfront reactor development and demonstration costs and time to deployment.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

DOE OSTI2022

Entropy–Preserving and Entropy–Stable Relaxation IMEX and Multirate Time–Stepping Methods

In this work, we propose entropy-preserving and entropy-stable partitioned Runge–Kutta (RK) methods. In particular, we extend the explicit relaxation Runge–Kutta methods to IMEX–RK methods and a class of explicit second-order multirate methods for stiff problems arising from scale-separable or grid-induced stiffness in a system. The proposed approaches not only mitigate system stiffness but also fully support entropy-preserving and entropy-stability properties at a discrete level. The key idea of the relaxation approach is to adjust the step completion with a relaxation parameter so that the time-adjusted solution satisfies the entropy condition at a discrete level. The relaxation parameter is computed by solving a scalar nonlinear equation at each timestep in general; however, as for a quadratic entropy function, we theoretically derive the explicit form of the relaxation parameter and numerically confirm that the relaxation parameter works the Burgers equation. Several numerical results for ordinary differential equations and the Burgers equation are presented to demonstrate the entropy-conserving/stable behavior of these methods. We also compare the relaxation approach and the incremental direction technique for the Burgers equation with and without a limiter in the presence of shocks.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Anisotropic Hot Spot Formation at a Grain Boundary in Shock-Compressed TATB High Explosive Crystal

Secondary high explosives (HEs) exhibit rich microstructure that promotes the formation of hot spots responsible for detonation initiation, but the role of microstructural interfaces remains poorly quantified. To this end, we develop extensions for the generalized crystal-cutting method (GCCM) to prepare molecular dynamics (MD) simulation cells containing grain boundaries (GBs) and other crystal–crystal interfaces with prescribed tilt and twist orientations. Using the GCCM, we perform MD simulations of shock interactions with a GB between the (001) and (100) crystal facets in the secondary HE TATB (1,3,5-triamino-2,4,6-trinitrobenzene). Our MD simulations reveal a strong directional dependence to the formation of a hot spot at the GB interface. In particular, transmission of the shock from the (001) grain to the (100) grain yields a hot spot in the (100) grain at the GB interface, whereas no hot spot is produced when an equivalent shock transits the GB in the opposite direction. We trace the origin of this GB anisotropy to three dominant factors: (1) the intrinsic differences in shock-deformation mechanisms and wave structures for the bulk (100) and (001) grains, which leads to distinct geometries and mechanical impedances upon shock arrival to the GB depending on which grains are donor or acceptor for the transmitted shock; (2) the different time intervals separating the initial shock rise and the formation of steady wave structures in the respective donor–acceptor configurations; and (3) the differences in time scales required to re-establish local thermal equilibrium. Interfacial hot spots form when these factors combine to impede development of a steady two-wave structure and instead induce a localized, pseudosingly shocked region that undergoes a higher rate of work production (resulting in a higher temperature) compared to when the steady two-wave structure develops further from the interface. The extensions to the GCCM approach presented here are anticipated to facilitate a wide range of MD studies that focus on understanding the role of crystal–crystal interfaces in molecular materials.

organic↗

DOE OSTI2022

Local analysis of fast magnetic reconnection

Fast magnetic reconnection is defined by the topology of the magnetic field lines changing on a timescale that is approximately an order of magnitude longer than the topology-conserving ideal-evolution timescale. Fast reconnection is an intrinsic property of Faraday's law when the evolving magnetic field depends non-trivially on all three spatial coordinates and is commonly observed—even when the effects that allow topology breaking are arbitrarily small. The associated current density need only be enhanced by a factor of approximately ten and flows in thin but broad ribbons along the magnetic field. These results follow from the variation in the separation of neighboring pairs of magnetic field lines, which in an ideal evolution typically increases exponentially with time, and the existence of a spatial scale below which magnetic field lines freely change their identities due to non-ideal effects, such as resistivity. Traditional reconnection theory ignores exponentially large variations and relies on the current density reaching a magnitude that is exponentially larger than is actually required. Here, an analysis of the behavior of magnetic field lines in the neighborhood of an arbitrarily chosen line is used to obtain more precise and rigorous results on intrinsic reconnection. The maximum parallel kinetic energy of collisionless charged particles is shown to have an exponential increase in time during a generic magnetic evolution.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2021

As a Matter of Tension: Kinetic Energy Spectra in MHD Turbulence

While magnetized turbulence is ubiquitous in many astrophysical and terrestrial systems, our understanding of even the simplest physical description of this phenomena, ideal magnetohydrodynamic (MHD) turbulence, remains substantially incomplete. As such, in this work we highlight the shortcomings of existing theoretical and phenomenological descriptions of MHD turbulence that focus on the joint (kinetic and magnetic) energy fluxes and spectra by demonstrating that treating these quantities separately enables fundamental insights into the dynamics of MHD turbulence. This is accomplished through the analysis of the scale-wise energy transfer over time within an implicit large eddy simulation of subsonic, super-Alfvénic MHD turbulence. Our key finding is that the kinetic energy spectrum develops a scaling of approximately k –4/3 in the stationary regime as magnetic tension mediates large-scale kinetic to magnetic energy conversion and significantly suppresses the kinetic energy cascade. This motivates a reevaluation of existing MHD turbulence theories with respect to a more differentiated modeling of the energy fluxes.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1978

On the attitude motion of an orbiting rigid body under the influence of gravity gradient torque

An investigation is made of the rotational motion of a rigid body orbiting the earth, under the influence of the geogravity-gradient torque. The attitude dynamics are formulated as perturbations from a nominal case. The perturbed equations are solved asymptotically, by the multiple scales technique. The independent variable, time, is extended into a space of higher dimension by means of new scales, fast, slow, etc. Integration of the equations is carried out separately in the new variables. The rapid and slow aspects of the attitude dynamics are systematically separated, resulting in a more efficient computer implementation and enhanced physical insight. The theory is applied to predict the attitude dynamics of an asymmetric rigid body satellite. A comparison is made of the maximum errors as the step size increases as predicted by the multiple scales solution and direct numerical integration. An improvement in computational speed of an order of magnitude is demonstrated.

Tao, Y. C.↗

NASA NTRS1986

EXOSAT observations of 3C 371

The N galaxy/BL Lacertae object 3C 371 was observed by EXOSAT on two occasions separated by 18 days in September 1984. The low-energy and medium-energy X-ray flux has varied on the time scale of years (in comparison with HEAO 1 and Einstein Observatory measurements), 18 days (between the authors' two observations), and 8 hr to about 25 minutes within the authors first observation. The 0.05-6 keV energy spectrum was determined using data acquired during the first observation with the low-energy telescope and channeltron matrix array detector (LE) and medium-energy proportional counters (ME). It is found that the spectrum probably consists of two components, with the steeper one at low photon energies.

Staubert, R.↗

DOE OSTI2022

Computational approaches to model X-ray photon correlation spectroscopy from molecular dynamics

X-ray photon correlation spectroscopy (XPCS) allows for the resolution of dynamic processes within a material across a wide range of length and time scales. X-ray speckle visibility spectroscopy (XSVS) is a related method that uses a single diffraction pattern to probe ultrafast dynamics. Interpretation of the XPCS and XSVS data in terms of underlying physical processes is necessary to establish the connection between the macroscopic responses and the microstructural dynamics. To aid the interpretation of the XPCS and XSVS data, we present a computational framework to model these experiments by computing the X-ray scattering intensity directly from the atomic positions obtained from molecular dynamics simulations. We compare the efficiency and accuracy of two alternative computational methods: the direct method computing the intensity at each diffraction vector separately, and a method based on fast Fourier transform that computes the intensities at all diffraction vectors at once. Furthermore, the computed X-ray speckle patterns capture the density fluctuations over a range of length and time scales and are shown to reproduce the known properties and relations of experimental XPCS and XSVS for liquids.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Biotechnological solutions for critical mineral recovery from unconventional feedstocks

Secure and sustainable metal recovery from unconventional feedstocks is needed to meet the mineral demands of energy, defense, and electronic technologies. Here, we highlight the potential to leverage nature’s ability to extract and differentiate metal ions in biotechnologies that could become the next generation of mining and refining. We describe bulk and trace processes and then discuss the advances and opportunities of two key bioprocesses: microbially mediated solubilization of metal ions from solid matrices (termed ‘bioleaching’) and bio-based separation of solubilized ions via selective adsorption to proteins. Both biotechnologies have advantages such as reduced energy input for leaching low-grade feedstocks and reduced organic solvent demand for separating ions with similar physiochemical properties but require more development for industrial scale recovery from unconventional feedstocks. Innovation in biological science and engineering may bring timely solutions to key challenges toward recovering critical minerals from unconventional feedstocks.

organic↗

DOE OSTI2024

Excitons in Hematite Fe 2 O 3 : Short-Time Dynamics from TD-DFT and Non-Adiabatic Dynamics Theories

We present a first-principles study of the short-time dynamics of excitons in hematite Fe 2 O 3 . We used time-dependent density functional theory (TD-DFT) with an underlying DFT+U treatment of electron interactions to characterize the electronic structure of excitons and nonadiabatic molecular dynamics theory (NA-MD) to determine their recombination (electronic ground-state recovery) and relaxation dynamics. Decoherence-corrected trajectory surface hopping approaches in NA-MD simulations yielded recovery times of ~1.1 to 1.8 ns and “higher-lying” exciton relaxation times of ~60 to 70 fs, in accord with experimentally derived lifetimes. With hematite phonons in the range of ~100 to 700 cm –1 , higher-lying excitons relax within one or two oscillations of the phonons before getting trapped into an electron–hole pair Exc-3 structure on the first excited state potential energy surface. This structure resembles already a pair of polarons (electron polaron plus hole polaron) with associated lattice distortions three (3) basal planes away. On longer time scales, the electron–hole bipolaronic pair hops to structures Exc-5, then Exc-7, then Exc-9, ... with the electron polaron and hole polaron separated by 5, 7, 9, ... basal planes in a process of charge separation. The largest frequency phonon ~672 cm –1 for the Exc-3 exciton structure is associated with the electron polaron moiety of the exciton. This phonon is a good candidate for giving rise to the recently observed and reported postexcitation transient IR absorption peak.

36 MATERIALS SCIENCE↗

NASA NTRS1993

Differentiating the role of land surface variability and cloudiness variability on global energy transport within the atmosphere and oceans

The following provides the final report on NASA IDP Project NAGW-1840 'Differentiating the Role of Land Surface Variability on Global Energy Transport within the Atmosphere and Oceans'. The project was designed to investigate the role of regional perturbations in the earth radiation budget on atmospheric and oceanic energy transports on an interannual basis. We proposed a modeling strategy based on an entropy extremum principle that could be used to separate the transports into oceanic and atmospheric components so as to better understand the effects of regional perturbations at the distinct atmospheric and oceanic time scales. The original focus was to consider the maintenance and year-to-year modulation of a large-scale, low-latitude North African-West Pacific Ocean net radiation dipole, which we had detected in the Nimbus 6 and 7 record of earth radiation budget measurements, and which necessitated significant cross-meridional energy transports to maintain global equilibrium. In addition, perturbations in the radiation balance term were to be partitioned into cloud-induced and surface-induced components to better understand the feedbacks between clouds and surface boundary conditions on interannual variability of the radiation balance.

Smith, Eric A.↗

DOE OSTI2022

On the Jet–Ejecta Interaction in 3D GRMHD Simulations of a Binary Neutron Star Merger Aftermath

Short γ-ray burst (sGRB) jets form in the aftermath of a neutron star merger, drill through disk winds and dynamical ejecta, and extend over four to five orders of magnitude in distance before breaking out of the ejecta. We present the first 3D general-relativistic magnetohydrodynamic sGRB simulations to span this enormous scale separation. They feature three possible outcomes: jet+cocoon, cocoon, and neither. Typical sGRB jets break out of the dynamical ejecta if (i) the bound ejecta's isotropic equivalent mass along the pole at the time of the BH formation is ≲10 -4 M ⊙ , setting a limit on the delay time between the merger and BH formation, otherwise, the jets perish inside the ejecta and leave the jet-inflated cocoon to power a low-luminosity sGRB; (ii) the postmerger remnant disk contains a strong large-scale vertical magnetic field, ≳1015 G; and (iii) if the jets are weak (≲10 50 erg), the ejecta's isotropic equivalent mass along the pole must be small (≲10 -2 M ⊙ ). Generally, the jet structure is shaped by the early interaction with disk winds rather than the dynamical ejecta. As long as our jets break out of the ejecta, they retain a significant magnetization (≲1), suggesting that magnetic reconnection is a fundamental property of sGRB emission. The angular structure of the outflow isotropic equivalent energy after breakout consistently features a flat core followed by a steep power-law distribution (slope ≳3), similar to hydrodynamic jets. In the cocoon-only outcome, the dynamical ejecta broadens the outflow angular distribution and flattens it (slope ~1.5).

Jets↗

NASA NTRS1994

Large-scale structure effects on the gravitational lens image positions and time delay

We compute the fluctuations in gravitational lens image positions and time delay caused by large-scale structure correlations. We show that these fluctuations can be expressed as a simple integral over the density power spectrum. Using the Cosmic Background Explorer (COBE) normalization we find that positions of objects at cosmological distances are expected to deviate from their true positions by few arcminutes. These deflections are not directly observable. The positions of the images relative to one another fluctuate by a few percent of the relative separation, implying that one does not expect multiple images to be produced by large-scale structure. Nevertheless, the fluctuations are larger than the observational errors on the positions and affect reconstructions of the lens potential. The time delay fluctuations have a geometrical and a gravitational contribution. Both are much larger than the expected time delay from the primary lens, but partially cancel each other. We find that large-scale structure weakly affects the time delay and time delay measurements can be used as a probe of the distance scale in the universe.

Seljak, Uros↗

DOE OSTI2020

Efficient Mesofluidic Separation of Large Particles in nuclear slurries - 20408

Efficient and effective particle separation is essential to cleanup of many nuclear wastes. For example, particle separation may be used to accelerate settle-decant bottlenecks that limit the throughput necessary to achieve the U.S. Department of Energy waste cleanup mission in a timely manner. Particle separation may be used to protect downstream processing equipment from pulses of solids that may be encountered near (within a few feet of) liquid-solid interfaces during waste transfer operations, thereby preventing work stoppages. Effective particle separation may permit efficient sludge washing. These are only a few of the ways in which particle separation is important. Yet, efficient and effective particle separation in nuclear processing environments remains challenging. For example, settle-decant operations permit larger and heavier particles to settle, leaving smaller, lighter particles suspended. However, settle-decant operations are slow, convection currents may resuspend solids, hindered settling of multicomponent slurries remains incompletely understood, and predictive models for settling in graduated cylinders fail to match limited observations of settling in large waste tanks. Additionally, pumping operations can cause turbulent resuspension of particles when the pump intake is close to the settled solids layer. Filtration techniques that use membranes or partially permeable barriers retain larger particles, permitting smaller particles to permeate. However, filters, including dead-end filtration, are prone to clogging and caking, operate at elevated pressures due to minimal void volume, and increase in pressure during operations, which require more control systems. Therefore, the need for high throughput particle separation techniques that operate with modest pressure drops persists. A novel mesofluidic separator presents the opportunity to effectively and efficiently accelerate the waste cleanup mission. The separator separates large particles from process streams across a broad range of particle sizes and has no moving parts or media to replace, regenerate, or clean. This separator design has an unusually large void volume, permitting operation at much higher flow rates (and lower pressures) than traditional filtration (e.g., dead-end filtration). Industrial-scale flow rates have been demonstrated. In performance testing, mesofluidic separators operate at might flow rates (>90 gpm (0.0006 m{sup 3}/s) in piping 3 inches (0.08 m) in diameter; Re>10{sup 5}), with modest pressure drop (∼25 psi (170 KPa) in testing). In complex, aggregating waste simulants, the separator loses <25% of flow projected over months without back pulsing or chemical cleaning. Mesofluidic separation presents infrastructure advantages, reduces risk, and provides mission impact. Infrastructure advantages include implementation within existing transfer systems, plug and play without facility modifications to safety systems, and operation at low pressures. The separator is flexible in location, may be positioned within or outside of waste tanks, and may replace or augment dead-end and cross-flow filters. Mesofluidic separation reduces risk by minimizing waste (no media to replace or regenerate and no cleaning chemicals to handle and dispose) and by minimizing filter change-out consistent with as low as reasonably achievable (ALARA) exposure to workers. The potential mission impact of these separators is substantial. The separator has potential to support, simplify, and accelerate in-farm transfers and waste feed delivery. Furthermore, the separator can free up settling and holding tanks in the Direct Feed Low-Activity Waste (DFLAW) mission by reducing or eliminating post-transfer settling and wait times with the potential to free up 1 Mgal (4000 m{sup 3}) of double-shell tank space. This paper quantitatively compares mesofluidic separation to dead-end filtration, discusses scale-up results, and considers the separator's potential to efficiently and effectively reduce the long-term environmental ability of particulate-rich nuclear wastes. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

DOE OSTI2024

QoS-aware edge AI placement and scheduling with multiple implementations in FaaS-based edge computing

Resource constraints on the computing continuum require that we make smart decisions for serving AI-based services at the network edge. AI-based services typically have multiple implementations (e.g., image classification implementations include SqueezeNet, DenseNet, and others) with varying trade-offs (e.g., latency and accuracy). The question then is how should AI-based services be placed across Function-as-a-Service (FaaS) based edge computing systems in order to maximize total Quality-of-Service (QoS). To address this question, we propose a problem that jointly aims to solve (i) edge AI service placement and (ii) request scheduling. These are done across two time-scales (one for placement and one for scheduling). Here we first cast the problem as an integer linear program. We then decompose the problem into separate placement and scheduling subproblems and prove that both are NP-hard. We then propose a novel placement algorithm that places services while considering device-to-device communication across edge clouds to offload requests to one another. Our results show that the proposed placement algorithm is able to outperform a state-of-the-art placement algorithm for AI-based services, and other baseline heuristics, with regard to maximizing total QoS. Additionally, we present a federated learning-based framework, FLIES, to predict the future incoming service requests and their QoS requirements. Our results also show that our FLIES algorithm is able to outperform a standard decentralized learning baseline for predicting incoming requests and show comparable predictive performance when compared to centralized training.

97 MATHEMATICS AND COMPUTING↗