Home Projects

Automated Solver Selection for Nuclear Simulations

RNET in collaboration with University of Oregon (Dr. Boyana Norris) is developing an add-in feature for the NEAMS (Nuclear Energy Advanced Modeling and Simulation) toolkit being developed by the Department of Energy. This feature will provide machine learning based solver selection capabilities for nuclear engineering simulations.

There exists several choices for solving the numerical sub problems occurring at various stages of nuclear reactor simulations. For example, linear solvers typically consume largest amount of time in transient simulations. As the linear systems produced by the transient simulations changes, the best preconditioned iterative solver also changes. Similarly, in applications with dynamic mesh adaptation, the mesh changes at runtime and the best linear solver could depend on the physical and geometric properties of the mesh. The right choice of the solvers during simulations will significantly contribute to the portability of the NEAMS toolkit. However, there is no governing theory for determining the best solution or the theory is too expensive to compute. Hence, the proposed add-in feature being developed by RNET will leverage machine learning techniques to automatically select the optimal solver based on run-time dependent features of the problem and the underlying compute architecture with minimal runtime overhead in solver selection.

Power Monitoring

Product Brief

RNET has been working on this DOE-funded Phase II project for the past two years. The objective of this project is to develop software/hardware tools that would perform “fine-grain” monitoring (i.e., millisecond monitoring) of power consumption when software code executes so that users can accurately know the amount power consumed by various parts of the software as it executes. Such fine-grain monitoring will be very helpful to DOE, Industry, and Academia in the context of “exa-scale” computing.

Currently, the power monitoring tools that are available can only perform “coarse-grain” monitoring (i.e., system level aggregate measurements on the order of seconds), in relation to the temporal granularity of performance measurements and the systems under test. These limitations do not allow application developers to realistically attempt to reduce power consumption of their applications. In this project, RNET is developing a low cost "fine-grained" component-level power monitoring hardware infrastructure plus relevant software and API that can be added to existing compute nodes or built into future motherboards. The power monitoring API will be embedded into some of commonly used performance monitoring tools (e.g., gprof, oprofile, HPC Toolkit, Eclipse, /proc, mpiP). These tools will allow transparent access to the application's power consumption at a hardware component level, allowing application developers to make informed performance/energy tradeoffs.


This project aims to improve the scalability and performance of CFD solvers of interest to the DoD through optimizations targeting emerging parallel architectures.

These systems will include multiple levels of parallelism that must be efficiently exploited, including clusters of multi-core processors, larger vector units (e.g., AVX) and application acceleration units (e.g., GPUs and Intel’s MIC). Optimizations include data structure reorganization to enable vectorization, loop optimizations to enable vectorization (through manual optimizations, semi-automatic source-to-source translations, and automatic compiler optimizations), GPU optimizations, and leveraging shared-memory on multi-core processing nodes. In addition, we are exploring using more advanced linear solvers (e.g., KSP solvers from the PETSc library such as GMRES) and fundamental modifications to the existing Gauss-Seidel solver that may improve the inherent scalability of the linear solver. In addition, coarse-grained parallelism optimizations include improved load balancing techniques.

Rad-Hard and ULP FPGA

RNET is currently working on two Phase II SBIR programs for NASA (NNX10CB47C and NNX12CA84C) in the development of field programmable gate array (FPGA) architectures that are highly tolerant to radiation.  NASA is interested in highly radiation tolerant devices for future exploration missions, including the Europa-Jupiter System Mission (EJSM) that requires a hardness of at least 2.9 Mrads total ionizing dose (TID).  Potential applications exist in space vehicles, satellites, orbiters, and others. 


RNET is working on a Phase II STTR titled “A MapReduce-like Data-Intensive Processing Framework for Native Data Storage and Formats” from the DOE (Department of Energy). The Ohio State University (OSU) is a collaborator on this STTR project. MapReduce is a very popular data analytic framework that is widely used in both industry and scientific research. Despite the popularity of MapReduce, there are several obstacles to applying it for developing some commercial and scientific data analysis applications.

This project will develop Native data FOrmat MapREDuce-like framework, iNFORMER, based on OSU’s SciMate architecture. The framework allows MapReduce-like applications to be executed over data stored in a native data format, without first loading the data into the framework. This addresses a major limitation of existing MapReduce-like implementations that require the data to be loaded into specialized file systems, e.g., the Hadoop Distributed File System (HDFS). The overheads and additional data management processes required for this translation can prevent MapReduce from being used in many commercial and scientific environments.


RNET Technologies has developed a video compression analysis system named Virtual Object Based Compression (VOBC) that includes many attractive options, e.g., real-time compression, AES encryption, streaming video mode, synchronized audio and video, object tracking, and 3-D stereo function capabilities. The compression performance of this new system is much better than that which can be achieved with MPEG-2 DVD technology and is comparable to that achieved by MPEG-4 AVC standard but at a relatively lower cost of implementation, both in hardware and software.

GPR for Plant Root Analysis

In February 2014 RNET was awarded a Phase I STTR titled “Ground Penetrating Radar System (GPR) and Algorithms for Fine Root Analysis” from the DOE (Department of Energy). The University of Dayton Research Institute (UDRI) is a collaborator on this STTR project.

GPR technology is a superlative choice for non-destructive imaging and analysis of roots. Current GPR-based root analysis techniques are designed for coarse root analysis and are unable to provide an accurate image of the root structure. The focus of this project is to develop an unconventional GPR system designed with greater accuracy, penetration depth, and imaging capabilities will extend the GPR technology for more accurate root analysis.

Radiation/Temperature Hardened Advanced Readout Array with Dynamic Power Modes

NASA has an interest in the development of advanced instruments and components for Lunar and planetary science missions.  Instrumentation is needed for the exploration of inner and outer planets and their moons, comets, asteroids, etc.  As a consequence, instrumentation systems must withstand the extreme surroundings experienced in space and planetary environments; radiation, temperature, pressure, launch/landing stresses, etc.  Specific areas related to instrument deployment for in situ sensors and sensor systems on a variety of space platforms including orbiters, flyby spacecraft, landers, rovers, balloons, and other aerial vehicles, sub-surface penetrators, and impactors are of interest. 

In the Phase I program,  an innovative digital readout integrated circuit (DROIC) architecture that will increase resolution and provide improved sensitivity will be developed.  The low power design will incorporate provisions to mitigate the effects of radiation and extreme temperatures. 

It is anticipated that the DROIC technology developed under this SBIR will be applicable to a variety of Flagship missions including EJSM, Mars Astrobiology Explorer-Cacher, Venus Climate Mission, and the Uranus Orbiter/probe.   Potential non-NASA applications include persistence surveillance applications for the military.  Commercial applications may exist in thermal imaging, hyperspectral imaging, or compressive sensing.

Fault Tolerant Mid-Wave Infrared (MWIR) Detector

Tactical and reconnaissance platforms rely on high performance, multifunctional optical sensor systems. At the heart of many of these optical sensor systems are mid-wave infrared (MWIR) focal plane arrays (FPA) for their ability to provide high resolution images regardless of the daylight conditions. Larger arrays are required to attain higher resolution and expanded capabilities.

Advanced ROIC Technology for Strained Layer Superlattice Photodetectors

The objective of this topic is to develop an optimized SLS-based Infrared FPA for high temperature operation with considerations for lower cost, size, weight, and power (C-SWAP).