SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Cray response to HPCS initiative. Model future petaflop computer interconnect. Parallel discrete event simulation techniques for large scale network simulation. Built on WarpIV engine. Run on laptop and Altix 3000. Can be sized up to 1000 simulated nodes per host node. Good parallel scaling characteristics. Flexible: multiple injectors, arbitration strategies, queue iterators, network topologies.
Topological study of digital computer programming efficiency in electric network analysis, considering cost of computation
Geomagnetic disturbance forecasting is based on the identification of solar wind structures and accurate determination of their magnetic field orientation. For nowcasting activities, this is currently a tedious and manual process. Focusing on the main driver of geomagnetic disturbances, the twisted internal magnetic field of interplanetary coronal mass ejections (ICMEs), we explore a convolutional neural network’s (CNN) ability to predict the embedded magnetic flux rope’s orientation once it has been identified from in situ solar wind observations. Our work uses CNNs trained with magnetic field vectors from analytical flux rope data. The simulated flux ropes span many possible spacecraft trajectories and flux rope orientations. We train CNNs first with full duration flux ropes and then again with partial duration flux ropes. The former provides us with a baseline of how well CNNs can predict flux rope orientation while the latter provides insights into real-time forecasting by exploring how accuracy is affected by percentage of flux rope observed. The process of casting the physics problem as a machine learning problem is discussed as well as the impacts of different factors on prediction accuracy such as flux rope fluctuations and different neural network topologies. Finally, results from evaluating the trained network against observed ICMEs from Wind during 1995–2015 are presented.
Because large unstructured datasets is important for many science domains, distributed graph analytics is critical to many scientists. Unfortunately, obtaining scaling and performance for irregular communication is challenging because contemporary network interconnects are primarily designed to maximize bandwidths of fixed-neighborhoods large-message exchanges (e.g., stencils). Although there is no consensus on the “best” network topologies for irregular communication, unstructured graph-based interconnects can be more suitable. We analyze three popular graph workloads – clustering, pattern enumeration, and traversal — on comparable networks (in terms of resources and costs) constructed from Jellyfish Random Regular, Dragonfly and Fat tree topologies, varying the routing algorithms. Using packet-level simulations, we demonstrate up to 60% improvement in communication time with Jellyfish due to diversity of the short paths between arbitrary endpoints, which can reduce overall network stalls and congestion.
The increasing adoption of distributed energy resources (DERs) over the last decade warrants a reconsideration of control of generation resources. This paper proposes a distributed Automatic Generation Control (AGC) using transmission-and-distribution (T&D) dynamic co-simulation framework for the efficient DPV frequency regulation services. The co-simulation framework allows AGC units to exchange the information for distributed AGC, based on their adopted communication network topology. As a result, a cost-effective automatic generation control is achieved with DPV and conventional generators. The proposed distributed AGC is based on the gossip algorithm in which the neighboring AGC units share the relevant local information with each other and updates their share of AGC regulation signal. Distributed photovoltaics (DPV) unit contribute to AGC response based on their headroom capacity via DER aggregators. The algorithm is tested on IEEE-14 bus transmission system under conditions of generation failure and random load variation to observe effective frequency regulations service offered by DPVs and other AGC units. The study shows that DPV can effectively participate in AGC with the proposed distributed control framework.
Network complexity and computational efficiency have become increasingly significant aspects of deep learning. Sparse deep learning addresses these challenges by recovering a sparse representation of the underlying target function by reducing heavily overparameterized deep neural networks. Specifically, deep neural architectures compressed via structured sparsity (e.g., node sparsity) provide low-latency inference, higher data throughput, and reduced energy consumption. In this article, we explore two well-established shrinkage techniques, Lasso and Horseshoe, for model compression in Bayesian neural networks (BNNs). To this end, we propose structurally sparse BNNs, which systematically prune excessive nodes with the following: 1) spike-and-slab group Lasso (SS-GL) and 2) SS group Horseshoe (SS-GHS) priors, and develop computationally tractable variational inference, including continuous relaxation of Bernoulli variables. We establish the contraction rates of the variational posterior of our proposed models as a function of the network topology, layerwise node cardinalities, and bounds on the network weights. Furthermore, we empirically demonstrate the competitive performance of our models compared with the baseline models in prediction accuracy, model compression, and inference latency.
Digital cellular mobile 'second generation' systems are now gradually being introduced into service; one such example is GSM, which will provide a digital voice and data service throughout Europe. Total coverage is not expected to be achieved until the mid '90's, which has resulted in several proposals for the integration of GSM with a geostationary satellite service. Unfortunately, because terrestrial and space systems have been designed to optimize their performance for their particular environment, integration between a satellite and terrestrial system is unlikely to develop further than the satellite providing a back-up service. This lack of system compatibility is now being addressed by system designers of third generation systems. The next generation of mobile systems, referred to as FPLMTS (future public land mobile telecommunication systems) by CCIR and UMTS (universal mobile telecommunication system) in European research programs, are intended to provide inexpensive, hand-held terminals that can operate in either satellite, cellular, or cordless environments. This poses several challenges for system designers, not least in terms of the choice of multiple access technique and power requirements. Satellite mobile services have been dominated by the geostationary orbital type. Recently, however, a number of low earth orbit configurations have been proposed, for example Iridium. These systems are likely to be fully operational by the turn of the century, in time for the implementation of FPLMTS. The developments in LEO mobile satellite service technology were recognized at WARC-92 with the allocation of specific frequency bands for 'big' LEO's, as well as a frequency allocation for FPLMTS which included a specific satellite allocation. When considering integrating a space service into the terrestrial network, LEO's certainly appear to have their attractions: they can provide global coverage, the round trip delay is of the order of tens of milliseconds, and good visibility to the satellite is usually possible. This has resulted in their detailed investigation in the European COST 227 program and in the work program of the European Telecommunications Standards Institute (ETSI). This paper will consider the system implications of integrating a LEO mobile service with a terrestrial service. Results will be presented from simulation software to show how a particular orbital configuration affects the performance of the system in terms of area coverage and visibility to a terminal for various locations and minimum elevation angle. Possible network topologies are then proposed for an integrated satellite/terrestrial network.
Analysis discusses STACOM (state criminal justic communication) network topology program used to design and evaluate digital telecommunications networks STACOM employs ESAU-WILLIAMS technique to search for direct links between system terminations and regional switching center. Inputs include traffic data, terminal locations, and functional requirements.
Dragonfly-based networks are an extensively deployed network topology in large-scale high-performance computing due to their cost-effectiveness and efficiency. The US will soon have three Exascale supercomputers for leadership class workloads deployed using dragonfly networks. Compared to indirect networks of similar scale, the dragonfly network has considerably reduced cable lengths, cable counts, and switch counts, resulting in significant network cost savings for a given system size, however, these cost reductions result in reduced global minimal paths and more challenging routing. Additionally, large scale dragonfly networks often require a taper at the global link level, resulting in less bisection bandwidth than is achievable in other traditional non-blocking topologies of equivalent scale. While dragonfly networks have been extensively studied, they have yet to be fully evaluated in an extreme scale (i.e., exascale) system that targets capability workloads. In this paper, we present the results of the first large scale evaluation of a dragonfly network on an exascale system (Frontier) and compare its behavior to a similar scale fat-tree network on a previous generation TOP500 system (Summit). This evaluation aims to determine the effect of network cost optimizations by measuring a tapered topology’s impact on capability workloads. Our evaluation is based on a collection of synthetic microbenchmarks, mini-apps, and full scale applications. It compares the scaling efficiencies of each benchmark between the dragonfly-based Frontier and the fat-tree-based Summit systems. Our results show that a dragonfly network is $\sim \mathbf{3 0 \%}$ more cost efficient than a fat-tree topology, which amortizes to $\sim 3 \%$ of an exascale system cost. Furthermore, while tapered dragonfly networks impose significant tradeoffs, the impacts are not as broad as initially thought and are mostly seen in applications with global communication patterns, particularly all-to-all (e.g., FFT-based algorithms), but also local communication patterns (e.g., nearest-neighbor algorithms) that are sensitive to network performance variability.
An inherent problem of conventional point-to-point wide area network (WAN) architectures is that they cannot translate optical transmission bandwidth into comparable user available throughput due to the limiting electronic processing speed of the switching nodes. The first solution to wavelength division multiplexing (WDM) based WAN networks that overcomes this limitation is presented. The proposed Lightnet architecture takes into account the idiosyncrasies of WDM switching/transmission leading to an efficient and pragmatic solution. The Lightnet architecture trades the ample WDM bandwidth for a reduction in the number of processing stages and a simplification of each switching stage, leading to drastically increased effective network throughputs. The principle of the Lightnet architecture is the construction and use of virtual topology networks, embedded in the original network in the wavelength domain. For this construction Lightnets utilize the new concept of lightpaths which constitute the links of the virtual topology. Lightpaths are all-optical, multihop, paths in the network that allow data to be switched through intermediate nodes using high throughput passive optical switches. The use of the virtual topologies and the associated switching design introduce a number of new ideas, which are discussed in detail.
With traditional event list techniques, evaluating a detailed discrete event simulation model can often require hours or even days of computation time. Parallel simulation mimics the interacting servers and queues of a real system by assigning each simulated entity to a processor. By eliminating the event list and maintaining only sufficient synchronization to insure causality, parallel simulation can potentially provide speedups that are linear in the number of processors. A set of shared memory experiments is presented using the Chandy-Misra distributed simulation algorithm to simulate networks of queues. Parameters include queueing network topology and routing probabilities, number of processors, and assignment of network nodes to processors. These experiments show that Chandy-Misra distributed simulation is a questionable alternative to sequential simulation of most queueing network models.
With traditional event-list techniques, evaluating a detailed discrete-event simulation-model can often require hours or even days of computation time. By eliminating the event list and maintaining only sufficient synchronization to ensure causality, parallel simulation can potentially provide speedups that are linear in the numbers of processors. A set of shared-memory experiments, using the Chandy-Misra distributed-simulation algorithm, to simulate networks of queues is presented. Parameters of the study include queueing network topology and routing probabilities, number of processors, and assignment of network nodes to processors. These experiments show that Chandy-Misra distributed simulation is a questionable alternative to sequential-simulation of most queueing network models.
The main contribution of the effort in the last two years is the introduction of the MOPPS system. After doing extensive literature search, we introduced the system which is described next. MOPPS employs a new solution to the problem of managing programs which solve scientific and engineering applications on a distributed processing environment. Autonomous computers cooperate efficiently in solving large scientific problems with this solution. MOPPS has the advantage of not assuming the presence of any particular network topology or configuration, computer architecture, or operating system. It imposes little overhead on network and processor resources while efficiently managing programs concurrently. The core of MOPPS is an intelligent program manager that builds a knowledge base of the execution performance of the parallel programs it is managing under various conditions. The manager applies this knowledge to improve the performance of future runs. The program manager learns from experience.
Predictive models of signaling pathways have proven to be difficult to develop. Reasons include the uncertainty in the number of species, the complexity in species’ interactions, and the sparseness and uncertainty in experimental data. Traditional approaches to developing mechanistic models rely on collecting experimental data and fitting a single model to that data. This approach works for simple systems but has proven unreliable for complex systems such as biological signaling networks. For example, uncertainty and sparseness of the data often result in overfitted models that have little predictive value beyond recapitulating the experimental data itself. Thus, there is a need to develop new approaches to create predictive mechanistic models of complex systems. However, to determine the effectiveness of any new algorithm, a baseline model is needed to test its performance. To meet this need, we developed a method for generating artificial synthetic networks that are reasonably realistic and thus can be treated as ground truth models. These synthetic models can then be used to generate synthetic data for developing and testing algorithms designed to recover the underlying network topology and associated parameters. Here, we describe a simple approach for generating synthetic signaling networks that can be used for this purpose.
Candidate data communication techniques are identified, including dedicated links, local buses, broadcast buses, multiplex buses, and mesh networks. The design methodology for mesh networks is then discussed, including network topology and node architecture. Several concepts of power distribution are reviewed, including current limiting and mesh networks for power. The technology issues of packaging, transmission media, and lightning are addressed, and, finally, the analysis tools developed to aid in the communication design process are described. There are special tools to analyze the reliability and connectivity of networks and more general reliability analysis tools for all types of systems.
A performance optimized implementation of Delay Tolerant Networking (DTN) with the capacity of gigabit-per second rates is developed for the International Space Station (ISS) and missions demanding large amounts of communications bandwidth. An overview of the High-rate Delay Tolerant Networking (HDTN) architecture and support for different convergence layers is provided. This paper then presents an overview of the testing and integration efforts to evaluate interoperability and capability in relevant environments. The first was interoperability testing with DTN Marshall Enterprise (DTNME) which resulted in near-gigabit per second data rates. This was followed by ISS emulation testing with the Software Development and Integration Laboratory (SDIL) at the Lyndon B. Johnson Space Center (JSC) and local testing based on the ISS DTN network topology. The local tests resulted in the discovery of potential sources of performance loss in the network and demonstrated near-gigabit rates between HDTN and DTNME.
Audio and video streaming across delay tolerant networks are relatively new phenomena. During the Apollo 11 mission, video and audio were streamed directly back to Earth using fully analog radios. This streaming capability atrophied over time due to the gradual conversion to digital electronics accompanied with higher resolutions causing the required bit rates to outpace communication link performance. Additionally, 21st century space systems face the new requirement of interconnectedness. Delay Tolerant Networking (DTN) attempts to solve this requirement by uniting traditional point to point links into a robust and dynamic network. However, In order to avoid system bottlenecks, the High-Rate Delay Tolerant Networking (HDTN) implementation focuses on performance-optimization of the standards. This work extends the functionality of HDTN by implementing audio and video streaming, with the goal of demonstrating the practical application of high definition media streaming across space networks. A series of network topologies were created including simple point to point links and multi-node multi-hop networks. Test media in the form of prerecorded and live footage was streamed across the network. A set of objective quality metrics were established in order to measure the stream quality. A lunar network was emulated using a mixture of embedded ARM platforms.
Wireless mesh networks are a network topology where all the nodes of a system are able to communicate with every other node in the network. This enables an adaptable network that is scalable and has the capability to self-repair and self-configure. The Modular Rapidly Manufactured Small Sat (MRMSS) Project is a small satellite project where we are developing a modular CubeSat architecture. One of the goals of the project is to develop a system that is quick and simple to integrate with a minimal amount of wiring involved. Wireless mesh networks are well suited for this configuration because of the self- configuring and self-repairing aspects of the network. This enables a satellite developer to add subsystem nodes to the network without the need for much hardware re-design. This paper will detail the background of wireless mesh networks, the advantages and limitations of using wireless mesh networks for space applications, and the technical progress of the wireless mesh network development of the MRMSS project.