SEARCH · Search NASA
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.
This paper describes a novel analog Vector Array Processor (VAP) that was designed for use in real-time and ultra-low power image compression applications. This custom CMOS processor is based architectually on the Vector Quantization (VQ) algorithm in image coding, and the hardware implementation fully exploits the inherent parallelism built-in the VQ algorithm.
The implementation of certain algorithms on parallel processing computing architectures can involve partitioning contiguous elements into a fixed number of groups, each of which is to be handled by a single processor. It is desired to find an assignment of elements to processors that minimizes the sum of the maximum workloads experienced at each stage. This problem can be viewed as a multi-objective network optimization problem. Polynomially-bounded algorithms are developed for the case of two stages, whereas the associated decision problem (for an arbitrary number of stages) is shown to be NP-complete. Heuristic procedures are therefore proposed and analyzed for the general problem. Computational experience with one of the exact problems, incorporating certain pruning rules, is presented with one of the exact problems. Empirical results also demonstrate that one of the heuristic procedures is especially effective in practice.
The use of the Cooley-Tukey algorithm for computing the l-d FFT lends itself to a particular matrix factorization which suggests direct implementation by linearly-connected systolic arrays. Here we present a new systolic architecture that embodies this algorithm. This implementation requires a smaller number of processors and a smaller number of memory cells than other recent implementations, as well as having all the advantages of systolic arrays. For the implementation of the decimation-in-frequency case, word-serial data input allows continuous real-time operation without the need of a serial-to-parallel conversion device. No control or data stream switching is necessary. Computer simulation of this architecture was done in the context of a 1024 point DFT with a fixed point processor, and CMOS processor implementation has started.
Least-squares (LS) estimation is a basic operation in many signal processing problems. Given y = Ax + v, where A is a m x n coefficient matrix, y is a m x 1 observation vector, and v is a m x 1 zero mean white noise vector, a simple least-squares solution is finding the estimated vector x which minimizes the norm of /Ax-y/. It is well known that for an ill-conditioned matrix A, solving least-squares problems by orthogonal triangular (QR) decomposition and back substitution has robust numerical properties under finite word length effect since 2-norm is preserved. Many fast algorithms have been proposed and applied to systolic arrays. Gentleman-Kung (1981) first presented the trianglular systolic array for a basic Givens reduction. McWhirter (1983) used this array structure to find the least-squares estimation errors. Then by geometric approach, several different systolic array realizations of the recursive least-squares estimation algorithms of Lee et al (1981) were derived by Kalson-Yao (1985). Basic QR decomposition algorithms are considered in this paper and it is found that under a one-row time updating situation, the Householder transformation degenerates to a simple Givens reduction. Next, an improved least-squares estimation algorithm is derived by considering a modified version of fast Givens reduction. From this approach, the basic relationship between Givens reduction and Modified-Gram-Schmidt transformation can easily be understood. This improved algorithm also has simpler computational and inter-cell connection complexities while compared with other known least-squares algorithms and is more realistic for systolic array implementation.
A specialized program was developed for flow cytometric list-mode data using an heirarchical tree method for identifying and enumerating individual subpopulations, the method of principal components for a two-dimensional display of 6-parameter data array, and a standard sorting algorithm for characterizing subpopulations. The program was tested against a published data set subjected to cluster analysis and experimental data sets from controlled flow cytometry experiments using a Coulter Electronics EPICS V Cell Sorter. A version of the program in compiled BASIC is usable on a 16-bit microcomputer with the MS-DOS operating system. It is specialized for 6 parameters and up to 20,000 cells. Its two-dimensional display of Euclidean distances reveals clusters clearly, as does its 1-dimensional display. The identified subpopulations can, in suitable experiments, be related to functional subpopulations of cells.
Spectral aerosol optical depths, surface hemispherical-directional reflectance factors, and bihemispherical reflectances (albedos) are retrieved for an area of Glacier National Park using spectral, multiangle imagery obtained with the airborne advanced solid state array spectroradiometer (ASAS). The retrieval algorithms are described and are identical in principle to those being devised for use by the multiangle imaging spectroradiometer (MISR) which will fly on the EOS-AMI spacecraft in 1998. As part of its science mission, MISR will produce global coverage of both aerosol amounts an an surface reflection properties. The results in this paper represent the initial effort in applying the MISR algorithms to real data. These algorithms will undergo additional testing and validation as more multiangle data become available.
When setting out to model and/or simulate a complex mechanical or electrical system, a modeler is faced with a vast array of tools, software, equations, algorithms and techniques that may individually or in concert aid in the development of the model. Mature requirements and a well understood purpose for the model may considerably shrink the field of possible tools and algorithms that will suit the modeling solution. Is the model intended to be used in an offline fashion or in real-time? On what platform does it need to execute? How long will the model be allowed to run before it outputs the desired parameters? What resolution is desired? Do the parameters need to be qualitative or quantitative? Is it more important to capture the physics or the function of the system in the model? Does the model need to produce simulated data? All these questions and more will drive the selection of the appropriate tools and algorithms, but the modeler must be diligent to bear in mind the final application throughout the modeling process to ensure the model meets its requirements without needless iterations of the design. The purpose of this paper is to describe the considerations and techniques used in the process of creating a functional fault model of a liquid hydrogen (LH2) system that will be used in a real-time environment to automatically detect and isolate failures.
Time-varying spacecraft magnetic fields or stray fields are a problem for magnetometer systems. While constant fields can be removed with zero offset calibration, stray fields are difficult to distinguish from ambient field variations. Putting two magnetometers on a long boom and solving for both the ambient and stray fields can be a good idea, but this gradiometer solution is even more susceptible to noise than a single magnetometer. Unless the stray fields are larger than the magnetometer noise, simply averaging the two measurements is a more accurate approach. If averaging is used, it may be worthwhile to explicitly estimate and remove stray fields. Models and estimation algorithms are provided for solar array, arcjet and reaction wheel fields.
The Multi-Anode Microchannel Array (MAMA) is a photon counting detector which utilizes a photocathode for photon to electron conversion, a microchannel plate (MCP) for signal amplification and a proximity focused anode array for position sensitivity. The detector electronics decode the position of an event through coincidence discrimination. The decoding algorithm which associates a given event with the appropriate pixel is determined by the geometry of the array. A new algorithm incorporated into a CMOS Application Specific Integrated Circuit (ASIC) decoder which improves the pixel spatial resolution is described. The new algorithm does not degrade the detector throughput and does not require any modifications to the detector tube. The standard MAMA detector has a pixel size of 25 x 25 square microns, but with the new decoder circuit the pixel size is reduced to 12.5 x 12.5 square microns. We have built the first set of decode electronics utilizing the new ASIC chips and report here on the first imaging tests of this system.
Algorithms for failure detection and identification for redundant noncolinear arrays of single degree of freedom gyros and accelerometers are described. These algorithms are optimum in the sense that detection occurs as soon as it is no longer possible to account for the instrument outputs as the outputs of good instruments operating within their noise tolerances, and identification occurs as soon as it is true that only a particular instrument failure could account for the actual instrument outputs within the noise tolerance of good instruments. An estimation algorithm is described which minimizes the maximum possible estimation error magnitude for the given set of instrument outputs. Monte Carlo simulation results are presented for the application of the algorithms to an inertial reference unit consisting of six gyros and six accelerometers in two alternate configurations.
Approximations for the power levels at the output of an adaptive array that uses the diagonally loaded sample-matrix inversion (SMI) algorithm are derived. Diagonal loading is a technique where the diagonal of the covariance matrix is augmented with a positive or negative constant prior to inversion. This paper examines how the signal-to-interference-plus-noise ratio (SINR) and signal-to-interference ratio (SIR) at the array output vary with the number of samples taken when the input signals are CW. It is shown that positive loading produces more rapid convergence with a reduction in output SIR. Negative loading provides an improved SIR level, but positive loading produces more rapid convergence. Negative loading provides an improved SIR level, but the output power levels are erratic and slow to converge. Simulation results which verify the theoretical procedure are given.
This paper describes two processing algorithms that can be implemented optically: the Radon transform and angular correlation. These two algorithms can be combined in one optical processor to extract all the basic geometric and amplitude features from objects embedded in video imagery. We show that the internal amplitude structure of objects is recovered by the Radon transform, which is a well-known result, but, in addition, we show simulation results that calculate angular correlation, a simple but unique algorithm that extracts object boundaries from suitably threshold images from which length, width, area, aspect ratio, and orientation can be derived. In addition to circumventing scale and rotation distortions, these simulations indicate that the features derived from the angular correlation algorithm are relatively insensitive to tracking shifts and image noise. Some optical architecture concepts, including one based on micro-optical lenslet arrays, have been developed to implement these algorithms. Simulation test and evaluation using simple synthetic object data will be described, including results of a study that uses object boundaries (derivable from angular correlation) to classify simple objects using a neural network.
Minimax algorithms for failure detection and identification for redundant noncolinear arrays of single-degree-of-freedom gyros and accelerometers are described. These algorithms are optimum in the sense that detection occurs as soon as it is no longer possible to account for the instrument outputs as the outputs of good instruments operating within their noise tolerances, and identification occurs as soon as it is true that only a particular instrument failure could account for the actual instrument outputs within the noise tolerance of good instruments. An estimation algorithm is described which minimizes the maximum possible estimation error magnitude for the given set of instrument outputs. Monte Carlo simulation results are presented for the application of the algorithms to an inertial reference unit consisting of six gyros and six accelerometers in two alternate configurations.
A set of language primitives designed to allow the specification of parallel numerical algorithms at a higher level is described. The authors focus on tensor product array computations, a simple but important class of numerical algorithms. They consider first the problem of programming one-dimensional kernel routines, such as parallel tridiagonal solvers, and then look at how such parallel kernels can be combined to form parallel tensor product algorithms.
Distributed memory architectures offer high levels of performance and flexibility, but have proven awkard to program. Current languages for nonshared memory architectures provide a relatively low level programming environment, and are poorly suited to modular programming, and to the construction of libraries. A set of language primitives designed to allow the specification of parallel numerical algorithms at a higher level is described. Tensor product array computations are focused on along with a simple but important class of numerical algorithms. The problem of programming 1-D kernal routines is focused on first, such as parallel tridiagonal solvers, and then how such parallel kernels can be combined to form parallel tensor product algorithms is examined.
This presentation presents the results of an algorithm to find the leading edge stagnation point using a sensor array with unknown calibration.
The potential of using a minicomputer/array-processor system for the efficient solution of large-scale, nonlinear, finite-element problems is studied. A Prime 750 is used as the host computer, and a software simulator residing on the Prime is employed to assess the performance of the Floating Point Systems AP-120B array processor. Major hardware characteristics of the system such as virtual memory and parallel and pipeline processing are reviewed, and the interplay between various hardware components is examined. Effective use of the minicomputer/array-processor system for nonlinear analysis requires the following: (1) proper selection of the computational procedure and the capability to vectorize the numerical algorithms; (2) reduction of input-output operations; and (3) overlapping host and array-processor operations. A detailed discussion is given of techniques to accomplish each of these tasks. Two benchmark problems with 1715 and 3230 degrees of freedom, respectively, are selected to measure the anticipated gain in speed obtained by using the proposed algorithms on the array processor.
A new approach to the design of multivariable control systems using the inverse Nyquist array method is proposed. The technique utilizes a conjugate direction function minimization algorithm to achieve dominance over a specified frequency range by minimizing the ratio of the moduli of the off-diagonal terms to the moduli of the diagonal term of the inverse open loop transfer function matrix. The technique is easily implemented in either a batch or interactive computer mode and will yield diagonalization when previously suggested methods fail. The proposed method has been successfully applied to design a control system for a sixteenth order state model of the F-100 turbofan engine with three inputs.