Search

The Larson–Miller parameter (LMP) offers an efficient and fast scheme to estimate the creep rupture life of alloy materials for high-temperature applications; however, poor generalizability and dependence on the constant C often result in sub-optimal performance. In this work, we show that the direct rupture life parameterization without intermediate LMP parameterization, using a gradient boosting algorithm, can be used to train ML models for very accurate prediction of rupture life in a variety of alloys (Pearson correlation coefficient >0.9 for 9–12% Cr and >0.8 for austenitic stainless steels). In addition, the Shapley value was used to quantify feature importance, making the model interpretable by identifying the effect of various features on the model performance. Finally, a variational autoencoder-based generative model was built by conditioning on the experimental dataset to sample hypothetical synthetic candidate alloys from the learnt joint distribution not existing in both 9–12% Cr ferritic–martensitic alloys and austenitic stainless steel datasets.

The objective of traditional phasor measurement unit (PMU) placement approaches is to find the minimal number of PMUs to achieve full system observability. Very few consider the probability and the consequence severity of observability loss. For example, following cascading failures, lines are tripped, and the topology of the grid changes, which may result in observability loss for parts of the grid. With limited number of PMUs, it is desired that the PMUs should still sustain observability for the power grid after top-severe cascading failures, other than after those cascading failures of low severity. To address the above issues related with the consequence severity of observability loss, this study proposes a risk-oriented PMU placement approach. The proposed approach uses two indices, namely the probability of observability loss and the severity of its corresponding consequence to design the optimal PMU placement strategy. The authors show that the proposed approach is able to capture the critical events, while traditional approaches fail to do so. Extensive simulation results carried out on the IEEE 39-bus and 118-bus systems demonstrate the effectiveness of the proposed risk-oriented approach.

Buildings are a primary consumer of electricity in the United States and are also increasingly being perceived as providers of grid services such as load shifting, shedding and modulation. High fidelity models of building electricity consumption are needed to set appropriate baselines for measurement and verification (M&V) of controllers designed for energy efficient operation of buildings and to enable buildings to provide grid services via. participation in demand response programs. State-of-the-art modeling techniques for building electricity consumption either rely on physics-based models, or extensive instrumentation of the building envelope to gather “big” data to train machine learning based models such as deep neural networks. While physics-based models are often limited by their accuracy, it is not always feasible to gather a significant amount of field data required to train machine learning based models with sufficient accuracy. In this paper, we explore the use of transfer learning-based strategies to address unsatisfactory accuracy of models for estimating building electricity consumption when available field data for training is sparse or of unacceptable quality. In particular, we transfer knowledge in the form of data and parameters, from physics-based simulation frameworks to the field to improve the model accuracy, thus resulting in a physics-informed machine learning framework. We evaluated the efficacy of our approach on field data collected from six commercial buildings and our results indicate that the proposed transfer learning-based models provide comparative (and in some cases better) accuracy than state-of-the-art machine learning and deep learning solutions, with just one month of field data.

It is well-known that delays and disruptions that originate at certain airports end up spreading spread on the network to other airports, sometimes straining a large fraction of the system. Hence data-driven modeling of this phenomenon is important from the standpoint of planning and strategy. In this work, utilizing the publicly available data from the Federal Aviation Administration (FAA), we employ the well-known influence maximization paradigm, to mine for the top airports in the US airport network, from the perspective of delay spread. In the process, we develop a diffusion simulator and implement a greedy algorithm to aid our task. We then explore the temporal trends by considering one winter month and one summer month in our analysis. Finally, we characterize the clusters in the US airport network and investigate the patterns of delay spread when airports in each of the clusters are affected.

Cost-efficient selection and scheduling of a subset of geographically distributed resources to meet the demands of a scientific workflow is a challenging problem. The problem is exacerbated by uncertainties in demand and availability of resources. In this paper, we present a stochastic optimization based framework for robust decision making in the selection of distributed resources over a planning horizon under demand uncertainty. We present a novel two-stage stochastic programming model for resource selection, and implement an L-shaped decomposition algorithm to solve this model. A Sample Average Approximation algorithm is integrated to enable stochastic optimization to solve problems with a large number of scenarios. Using the metric of stochastic solution, we demonstrate up to 30% cost reduction relative to solutions without explicit consideration of demand uncertainty for a 24-month problem. We also demonstrate up to 54% cost reduction relative to a previously developed solution for a 36-month problem. We further argue that the composition of resources selected is superior to solutions computed without explicit consideration of uncertainties. Given the importance of resource selection and scheduling of complex scientific workflows, especially in the context of commercial cloud computing, we believe that our novel stochastic programming framework will benefit many researchers as well as users of distributed computing resources.

The awareness about software vulnerabilities is crucial to ensure effective cybersecurity practices, the development of high-quality software, and, ultimately, national security. This awareness can be better understood by studying the spread, structure and evolution of software vulnerability discussions across online communities. This work is the first to evaluate and contrast how discussions about software vulnerabilities spread on three social platforms-Twitter, GitHub, and Reddit. Moreover, we measure how user-level e.g., bot or not, and content-level characteristics e.g., vulnerability severity, post subjectivity, targeted operating systems as well as social network topology influence the rate of vulnerability discussion spread. To lay the groundwork, we present a novel fundamental framework for measuring information spread in multiple social platforms that identifies spread mechanisms and observables, units of information, and groups of measurements. We then contrast topologies for three social networks and analyze the effect of the network structure on the way discussions about vulnerabilities spread. We measure the scale and speed of the discussion spread to understand how far and how wide they go, how many users participate, and the duration of their spread. To demonstrate the awareness of more impactful vulnerabilities, a subset of our analysis focuses on vulnerabilities targeted during recent major cyber-attacks and those exploited by advanced persistent threat groups. One of our major findings is that most discussions start on GitHub not only before Twitter and Reddit, but even before a vulnerability is officially published. The severity of a vulnerability contributes to how much it spreads, especially on Twitter. Highly severe vulnerabilities have significantly deeper, broader and more viral discussion threads. When analyzing vulnerabilities in software products we found that different flavors of Linux received the highest discussion volume. We also observe that Twitter discussions started by humans have larger size, breadth, depth, adoption rate, lifetime, and structural virality compared to those started by bots. On Reddit, discussion threads of positive posts are larger, wider, and deeper than negative or neutral posts. We also found that all three networks have high modularity that encourages spread. However, the spread on GitHub is different from other networks, because GitHub is more dense, has stronger community structure and assortativity that enhances information diffusion. We anticipate the results of our analysis to not only increase the understanding of software vulnerability awareness but also inform the existing and new analytical frameworks for simulating information spread e.g., disinformation across multiple social environments online.

In many complex networked systems, such as online social networks, activity originates at certain nodes and subsequently spreads on the network through influence. In this work, we consider the problem of modeling the spread of influence and the identification of influential entities in a complex network when nodal activation can happen via two different mechanisms. The first mechanism of activation stems from factors that are intrinsic to the node. The second mechanism comes from the influence of connected neighbors. After introducing the model, we provide an algorithm to mine for the influential nodes in such a scenario by modifying the well-known influence maximization algorithm. We sketch a proof of the submodularity of the influence function under the new formulation and demonstrate the same on larger graphs. Based on the model, we explain how influential content creators can drive engagement on social media platforms. Using additional experiments on a Twitter dataset, we then show how the formulation can be applied to real-world social media datasets. Finally, we derive a centrality metric that takes into account both the mechanisms of activation and provides for an accurate, computationally efficient, alternate approach to the problem of identifying influencers under intrinsic activation.

Scheduling of complex scientific workflows on geographically distributed resources is a challenging problem. Selection and scheduling of a subset of available resources to meet a given demand in a cost efficient manner is the first step of this complex process. In this paper, we develop a method to compute cost-efficient selection and scheduling of resources under demand uncertainties. Building on the techniques of Sample Average Approximation and Genetic Algorithms, we demonstrate that our method can lead up to \(24\%\) improvement in costs when demand uncertainties are explicitly considered. We present the results from our preliminary work in the context of a high energy physics application, the Belle II experiments, and believe that the work will equally benefit other scientific workflows executed on distributed resources with demand uncertainties. The proposed method can also be extended to include uncertainties related to resource availability and network performance.

In this work we propose a novel formulation that models the attack and compromise on a cyber network as a combination of two parts — direct compromise of a host and the compromise occurring through the spread of the attack on the network from a compromised host. The model parameters for the nodes are a concise representation of the host profiles that can include the risky behaviors of the associated human users while the model parameters for the edges are based on the existence of vulnerabilities between each pair of connected hosts. The edge models relate to the summary representations of the corresponding attack-graphs. This results in a formulation based on Random Walk with Restart (RWR) and the resulting centrality metric can be solved for in an efficient manner through the use of sparse linear solvers. Thus the formulation goes beyond mere topological considerations in centrality computations by summarizing the host profiles and the attack graphs into the model parameters. The computational efficiency of the method also allows us to also quantify the uncertainty in the centrality measure through Monte Carlo analysis.

Spectral domain statistical analysis of RF circuits, combining a circuit simulator, which models the circuit part and a full-wave field solver, which models the passive elements, is presented in this paper. This paper first illustrates the importance of the knowledge of correlation information in accurately modeling the probability density functions (PDFs) of eventual objective functions using a simple transmission line paradigm. Next, this paper looks at the statistical study of on-chip RF passives using the spiral inductor as an example. It is shown that larger process variations necessitate modeling by means of a quadratic response surface to preserve accuracy. This results in nonindependent non-Gaussian nonclosed-form PDFs for the equivalent-circuit parameters of the passives. This paper then proposes a hierarchical technique to perform statistical analysis of RF circuits based on y-parameter representation for the circuit and the passive element parts. The proposed technique obviates the need for optimization steps to derive the equivalent-circuit parameters for the electromagnetic objects and the need to compute the correlation matrix between the circuit equivalent elements, while maintaining accuracy. The proposed approach is illustrated for the statistical analysis of an RF amplifier and its differential version operating at 15.78 GHz. PDFs of various quantities of interest are derived and yield measures are computed.

We carefully revisit our definition of a social media signal from previous work both in terms of time-varying features within the data and the networked nature of the medium. Further, we detail our analysis of global patterns in Twitter over the month of June 2014, detect and categorize events, and illustrate how these analyses can be used to inform graph-based models of Twitter, namely using a recent network influence model called PhySense: similar to PageRank but tuned to behavioral analysis by leveraging a sociologically inspired probabilistic model. We ultimately identify a signature of information dissemination via analysis of time series and dynamic graph spectra and corroborate these findings through manual investigation of the data as a requisite step in modeling the diffusion process with PhySense. We have made our time series and dynamic graph analytical code available via a GitHub repository 1 and our data are available upon request.

The explosion in social media adoption has opened up new opportunities for next-generation personalized web and information exchange in big data scenarios. Making sense of the massive number of overlapping streams of information generated by hundreds of millions of users on large social networks requires novel analytics and scalable computational techniques. This paper introduces PHYSENSE, a scalable framework for influence computation and activity prediction on large online social networks (OSNs). Drawing inspiration from the Friedkin-Johnsen model for opinion change, PHYSENSE estimates and sets up sociological influence models to compute the diffusion of activity potential in the neighborhood of each of the nodes. PHYSENSE then scales these to significant parts of the entire OSN by propagating these activity potentials through an equivalent Helmholtz Green's function. Examples to show the enhanced quality of PHYSENSE in influence detection over popular existing methods based on variations of PageRank are presented. Additionally, for enhanced speedup, the community structures found in the social graphs along with low-rank updates are exploited in the acceleration of both the setup and the dynamic update phases of the influence computation.

This paper introduces PHYSENSE, a scalable framework for topic-dependent influence computation on large online social networks (OSNs) with application to generation of signatures of anomalous activity. PHYSENSE estimates and sets up sociological influence models to compute the diffusion of activity potential in the neighborhood of each of the nodes on the OSN. PHYSENSE then scales these to significant parts of the OSN by propagating the activity potentials through the graph topology, thereby generating the influence landscape in the form of an equivalent Green's function matrix. The computationally efficient dynamic update phase of PHYSENSE tracks the time and topic dependent changes in the influence landscape.

Full-wave simulation and design of interconnect systems is becoming increasingly important in view of the continuing trend towards miniaturization and dense integration of components in electronic systems. While full-wave simulation methods have advanced rapidly over the last decade, there is a growing need for superior automated design tools. In this paper we first motivate the need for better design space exploration in high dimensions and outline the key ingredients that are imperative for the development of a successful simulation-based automated design tool. In this direction we describe the application of Orthogonal Arrays as an efficient sampling scheme and the Gaussian Process Regression as a promising model generation scheme for high dimensional design space exploration. We also describe an adaptive strategy for the model refinement and illustrate the same with a full-wave multi-conductor transmission line example. Finally we conclude by outlining the open problems and research challenges in this direction.

The massive compute power offered by public clouds such as Amazon Web Services EC2 provides for a new paradigm in cost-effective and highly scalable parallel deployment of Electronic Design Automation (EDA) tools. Even though the advantages are myriad, customers perceive an inherent security risk in exposing their IP to the cloud. In this work, we start by outlining the shortcomings of established encryption techniques for use in public machines. Using the example of electromagnetic simulation, we show how the IP (layout and technology) may be reverse-engineered from the Green's function matrix by utilizing the Multi Dimensional Scaling approach. We then propose encryption schemes to defeat such sophisticated hacking attempts using principles of homomorphic encryption, while enabling scalability and computational benefits of public clouds

Synthesis of multi-GigaHertz radio frequency circuits brings together difficult challenges related to simulation, extraction and multidimensional space search. The standard approach of mapping all electromagnetic parasitics into parametric RLC models prior to synthesis is extremely restrictive especially when broadband and full-wave models with high accuracy are needed. In the presented approach, a two-stage macromodel that creates broadband, accurate parametric representations of passives, in particular spiral inductors, is developed. The broadband nature is captured through the Vector Fitting algorithm. The macromodels are implemented via efficient nonlinear, multidimensional regression using Relevance Vector Machine, and are coupled into circuit simulators through admittance parameters. Subsequently, optimization on both active and passive parameters are carried out simultaneously, thereby bypassing the ad hoc nature of two stage (actives and passives) approximate optimization. Two standard low-noise amplifier topologies are synthesized with tight performance constraints at center frequencies 5, 10 and 12 GigaHertz in order to demonstrate the frequency scalability of the methodology.

A parametric model of a generic differential ten-layer microprocessor package line with two motherboard layers has been developed. A dimensionality reduction scheme and a reusable, multi-dimensional look-up table precede the global optimization phase which is facilitated by a smooth interpolation scheme based on splines. An accelerated boundary element based full-wave electromagnetic solver has been used to construct a look-up table of S-parameters for a number of designs. The second phase features a custom local optimizer incorporating all the variables without any dimension reduction. This methodology has been applied to automated synthesis of the differential package line resulting in a significant improvement of the return loss performance.

High-speed applications in radio frequency, mixed-signal, and digital packaged electronics now necessitate the use of 3D, full-wave electromagnetic field solvers. The widely varied applications now involved in System-in-Package (SiP), System-on-Chip (SoC), Package-on-Package (PoP), three-dimensional integrated circuit (3DIC) technology including through-Silicon vias (TSVs) range from design and verification of the power delivery network (PDN), signal integrity and channel modeling of high-speed serial and parallel links, electromagnetic interference (EMI) and radiation, simultaneous switching noise (SSN), and voltage drop and ground bounce modeling amongst others. In addition, designers need to co-design chip I/O, packages, and board subsections nearly simultaneously in order to satisfy system-level signal integrity (SI), power integrity (PI), and EMI constraints. EM field solution is needed not merely in the complete verification of such systems but also in early design. Unfortunately, most EM simulators continue to be verification tools, merely masquerading as design tools. In this paper we discuss the challenges and needs associated with EM-aware design. In particular, we focus on digital microelectronics packages and RF frontend integrated circuits. We discuss how multidimensional parametrization, optimization, and synthesis need to be taken into account while creating simulator flows that will enable early design.

A geometry based parametric model of a differential high-speed line traversing a ten-layer microprocessor package is developed. This model is used to undertake a detailed study of the effect of the various geometrical parameters on the return loss performance of such a package. The forward problem is solved using a fast, full-wave electromagnetic (EM) solver. The effect of various types of routing in the intermediate layers is examined closely. A sensitivity analysis based on parametric sweeps is carried out to identify the key variables. A hierarchical response surface based optimization is carried out to arrive at an optimum structure. A global optimizer based on simulated annealing is harnessed to find the optimum of non-linear and, in general, non-convex functions. The optimized structure exhibits excellent return loss characteristics translating into higher channel bandwidth.

Modeling process variations in circuits is becoming necessary as the technology continues to shrink in size. When this is coupled with increasing frequencies of operation of circuits in communication systems (>10 GHz), the use of full-wave field solvers is necessary to model the on-chip passive components such as spiral inductors and transformers. Thus it becomes necessary to couple full-wave field solvers and SPICE-like simulators for accurate statistical analysis of RF circuits. The present paper reviews the issues associated with using circuit models for electromagnetically modeled passive elements in conjunction with SPICE for statistical analysis. It then proposes a method which circumvents the outlined limitations.

This paper presents a statistical study of the various parameters of interest of an on- chip spiral inductor in the 10 GHz range using the PMCHWT formulation [1]. Variability in the conductivity of the CMOS grade Silicon substrate and trace metal widths are assumed. We show that EM analysis of these inductors can accurately help in understanding the impact of process variations especially at high frequencies. The use of the PMCHWT formulation also permits analysis of the electrical properties of the substrate that are particularly important at higher clock and carrier frequencies. To enable fast extraction of statistics, a simple 2D Lagrange interpolation technique is used to approximate the surfaces of the quantities of interest and then a Montecarlo operation is carried out on the interpolated 2D polynomial. We extract the probability density functions (PDFs) of the effective resistance (R), the inductance (L) and the quality factor (Q) of a spiral on a lossy Silicon substrate.

A novel PBG cell based on micromachining of silicon using wet anisotropic etching has been considered. Since this is based on etching of the silicon substrate, it is amenable to fabrication with standard silicon processes and integration with millimeter wave circuits. We characterize this kind of PBG cell by full wave simulations using a time domain code. For the purpose of characterization, the scenario of a 50 ohm microstrip line placed on a silicon substrate which is anisotropically etched to create patterns with sloping walls is considered. This is shown to produce the well known PBG response of stop bands in certain frequency bands. We look at the variation in the transmission coefficient (S/sub 21/) response as the number of periods, length based average fill factor and depth of micromachining are varied. One application of a low pass filter has been proposed and simulated results are given.

Micromachined antennas are recieving great interest as carrier frequencies move higher into the frequency spectrum due to their superior performance and amenability for integration with active devices. However their design is cumbersome owing to the complexity of the structure. To overcome this, in this paper, an iterative procedure is suggested to facilitate fast design of micromachined patch antennas based on a simulation study. A microstrip line on a micromachined Silicon substrate is simulated in a full wave simulator by solving for the ports only. From the obtained propagation constant, the effective dilectric constant for the micromachined substrate is estimated. The process is repeated for a number of values of the width of the microstrip and a plot is made for the variation of the effective dielectric constant with the microstrip width. Then an iterative method in combination with the extrapolated permittivity which includes the effect of cavity extensions in all the directions, is used to obtain the width and the corresponding effective dielectric constant. This method has been verified to be quite accurate by comparison with full wave simulations and hence it can function as a good starting point for designers to design micromachined antennas.