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.
The fixed-point smoothing problem of sequentially correlated processes is considered. The filtering technique of such processes is extended to smoothing problems by simultaneously estimating the state and the accumulated contribution of process noise. Suboptimal filtering and smoothing are also discussed when data noise is regarded white in designing filters although it is actually sequentially correlated.-
The quantization error introduced by the Winograd Fourier transform algorithm (WFTA) when implemented in fixed-point arithmetic is studied and compared with that of the fast Fourier transform (FFT). The effect of ordering the computational modules and the relative contributions of data quantization error and coefficient quantization error are determined. In addition, the quantization error introduced by the Good-Winograd (GW) algorithm, which uses Good's prime-factor decomposition for the discrete Fourier transform (DFT) together with Winograd's short length DFT algorithms, is studied. Error introduced by the WFTA is, in all cases, worse than that of the FFT. In general, the WFTA requires one or two more bits for data representation to give an error similar to that of the FFT. Error introduced by the GW algorithm is approximately the same as that of the FFT.
Fixed point probability vector of ergodic /cyclic/ transition matrices by matrix inversion technique
Table lookup/interpolation function generation for fixed point digital computations, illustrating sine-cosine generator
Monte Carlo simulation of autonomous orbit determination has validated the use of an 18-bit NASA Standard Spacecraft Computer (NSSC) for the extended Kalman filter. Dimensionally consistent scales are chosen for all variables in the algorithm, such that nearly all of the onboard computation can be performed in single precision without matrix square root formulations. Allowable simplifications in algorithm implementation and practical means of ensuring convergence are verified for accuracies of a few km provided by star/vertical observations
This investigation is divided functionally into three different areas: (1) study of bounded state space, (2) nonlinear smoothing theory, and (3) system identification. (1) Study of bounded state space: necessary and sufficient conditions for an optimal control are obtained for a bounded state space optimal control problem. The difficulty of determining the so-called jump conditions is eliminated; however, the problem of determining the points where the response either enters or leaves the boundary still remains unsolved. (2) Nonlinear smoothing theory: nonlinear fixed-interval, fixed-point and fixed-lag smoothing of a random signal generated by a stochastic differential equation are investigated. Results on the asymptotic stability of a linear constant-parameter fixed-interval smoothing filter are obtained. (3) System identification: a particular stochastic modelling problem is solved. An Ito stochastic integral equation is used to mathematically model a black box having multiple inputs and multiple outputs. A new method for identifying system parameters is presented.
For binary digit array of N bits, use N noise sources to feed N nonlinear operators; each flip-flop in digit array is set by nonlinear operator to reflect whether amplitude of generator which feeds it is above or below mean value of generated noise. Fixed-point uniform distribution random number generation method can also be used to generate random numbers with other than uniform distribution.
Many problems in analysis are concerned with the dependence upon parameters of fixed points of maps. For contraction mappings, criteria are relatively easy to obtain and have been known for some time. In the study of solutions of functional differential equations, more general results were needed. It is the purpose of this paper to give a rather general fixed-point theorem for condensing maps depending on a parameter, to prove continuous dependence and to indicate how many of the previous results are special cases.
Analysis of the bit-level operations involved in the convolution realizing a nonrecursive digital filter leads to hardware designs of digital filters based on the operation of counting. Two distinct designs are outlined: the first one is capable of very high speed but is rather expensive; the second is quite slow but has the advantages of low cost and high flexibility. The basic designs considered utilize fixed-point representation for the data and filter coefficients. Variants allowing floating-point representation of the coefficients are also described.
In this paper a technique is developed for choosing programing forms and bit configurations for digital filters that minimize the quantization errors. The technique applies to digital filters operating in fixed-point arithmetic in either open-loop or closed-loop systems, and is implemented by a digital computer program that is based on a digital simulation of the system. As an output the program gives the programing form required for minimum quantization errors, the total bit configuration required in the filter, and the location of the binary decimal point at each quantizer within the filter.
The table look-up approach to pattern recognition has been used for 3 years at several research centers in a variety of applications. A new version has been developed which is faster, requires significantly less core memory, and retains full precision of the input data. The new version can be used on low-cost minicomputers having 32K words (16 bits each) of core memory and fixed-point arithmetic; no special-purpose hardware is required. An initial FORTRAN version of this system can classify an ERTS computer-compatible tape into 24 classes in less than 15 minutes.
Sufficient conditions are obtained for the existence of periodic motion in a class of autonomous nonlinear differential equations of order greater than two. The approach is based on the decomposition of an equation into a linear and a nonlinear part. The analysis relies on some basic ideas from linear analysis and geometry. Sufficient conditions for a periodic solution are derived by means of a general topological principle referred to as the torus principle. The existence of a periodic solution is concluded by an appropriate use of the Brouwer fixed-point theorem.
In this brief paper we examine a Riccati equation decomposition due to Reid and Lainiotis and apply the result to the continuous time-invariant linear filtering problem. Exploitation of the time-invariant structure leads to integration-free covariance recursions which are of use in covariance analyses and in filter implementations. A super-linearly convergent iterative solution to the algebraic Riccati equation (ARE) is developed. The resulting algorithm, arranged in a square-root form, is thought to be numerically stable and competitive with other ARE solution methods. Certain covariance relations that are relevant to the fixed-point and fixed-lag smoothing problems are also discussed.
A controller designed to guide an automated vehicle to a specified target without external intervention is described. The intended application is to the requirements of planetary exploration, where substantial autonomy is required because of the prohibitive time lags associated with closed-loop ground control. The guidance algorithm consists of a set of piecewise-linear control laws for velocity and steering commands, and is executable in real time with fixed-point arithmetic. The use of a previously-reported object tracking algorithm for the vision system to provide position feedback data is described. Test results of the control system on a breadboard rover at the Jet Propulsion Laboratory are included.
The topics considered are: (1) the base of the number system, (2) precision control, (3) number representation, (4) arithmetic operations, (5) other basic operations, (6) elementary functions, and (7) exception handling. The possibility of doing without fixed-point arithmetic is also mentioned. The specifications are intended to be entirely at the level of a programming language such as FORTRAN. The emphasis is on convenience and simplicity from the user's point of view. Conforming to such specifications would have obvious beneficial implications for the portability of numerical software, and for proving programs correct, as well as attempting to provide facilities which are most suitable for the user. The specifications are not complete in every detail, but it is intended that they be complete in spirit - some further details, especially syntatic details, would have to be provided, but the proposals are otherwise relatively complete.
There exists a number of mathematical procedures for designing discrete-time compensators. However, the digital implementation of these designs, with a microprocessor for example, has not received nearly as thorough an investigation. The finite-precision nature of the digital hardware makes it necessary to choose an algorithm (computational structure) that will perform 'well-enough' with regard to the initial objectives of the design. This paper describes a procedure for estimating the required fixed-point coefficient wordlength for any given computational structure for the implementation of a single-input single-output LOG design. The results are compared to the actual number of bits necessary to achieve a specified performance index.
There exist a number of mathematical procedures for designing discrete-time compensators. However, the digital implementation of these designs, with a microprocessor, for example, has not received nearly as thorough an investigation. The finite-precision nature of the digital hardware makes it necessary to choose a computational structure that will perform adequately with regard to the initial objectives of the design. This paper describes a procedure for estimating the required fixed-point coefficient wordlength for any given computational structure for the implementation of a single-input single-output LQG design. The results are compared to the actual number of bits necessary to achieve a specified performance index.
A method is outlined for evaluating discrete and background starlight at the Skylab wavelengths so as to derive the total brightness of zodiacal light at five sky positions. These sky positions are north celestial pole, south ecliptic pole, vernal equinox and two places near the north galactic pole. Pioneer 10/11 imaging photopolarimeters were used to periodically measure sky brightness and polarization in the blue and red at heliocentric distances beyond 1.002 AU. Mean zodiacal light, which is assumed to be also solar color in total light, is then estimated by applying the method to all fixed-point observations.