NASA Technical Reports Server (NTRS) 20110016574: Enhanced Software for Scheduling Space-Shuttle Processing

0.12 MB·English

by  NASA Technical Reports Server (NTRS)

[b/(b+1)] binary convolutional code (or plexity is possible. This can be done by for NASA’s Jet Propulsion Laboratory.
a short block code) with maximum free assigning the m = log2N output bits of Further information is contained in a TSP
Hamming distance (or minimum dis- the inner encoder alternately to the in- (see page 1)
tance). The interleaver (π) permutes the phase and quadrature components of In accordance with Public Law 96-517,
output of the outer encoder. The inter- N2QAM modulation. In this case, the the contractor has elected to retain title to this
leaved data enter the inner encoder, bit-rate-to-bandwidth ratio will be invention. Inquiries concerning rights for its
which implements a rate-(m/m) [rate-1] 2bm/(b+1). commercial use should be addressed to:
recursive convolutional code. The mout- The advantage of this generic design Intellectual Property group
put bits are then mapped to one symbol can be made more apparent by citing an JPL
that belongs to a 2m-level modulation. example of b = 3 for 16QAM, for which Mail Stop 202-233
Because the inner code does not have m = 2. In this example, the number of 4800 Oak Grove Drive
redundancy, it is useless by itself; how- transitions per state of the inner TCM is Pasadena, CA 91109
ever, the combination of the inner and only 4, which is only 1/32 of the corre- [phone]
outer codes with the interleaver results sponding number for the previous case. Refer to NPO-20878, volume and number
in very powerful code. For MQAM This work was done by Dariush Divsalar, of this NASA Tech Briefs issue, and the
where M = N2, further reduction in com- Sam Dolinar, and Fabrizio Pollara of Caltech page number.
Enhanced Software for Scheduling Space-Shuttle Processing
Prototype software has been upgraded.
John F. Kennedy Space Center, Florida
The Ground Processing Scheduling into a flexible system that supports mul- system was perhaps the most challeng-
System (GPSS) computer program is tiple-flow, multiple-site scheduling and ing aspect of the project. The proto-
used to develop streamlined schedules that retains the strengths of the proto- type utilized a Prolog-like query lan-
for the inspection, repair, and refurbish- type while incorporating improvements guage that was scanned, parsed, and
ment of space shuttles at Kennedy Space in maintainability, enhanceability, and executed in a C program. The query
Center. A scheduling computer pro- portability. The major enhancements code was problematic and difficult to
gram is needed because space-shuttle were the following: understand. The re-engineering in-
processing is complex and it is fre- • The implementation of container ob- volved the building of a real (but simi-
quently necessary to modify schedules to jects (e.g., lists and maps) was made lar) query language, utilizing the
accommodate unanticipated events, un- more efficient by use of the C++ Stan- FLEX language and the Bison pro-
availability of specialized personnel, un- dard Template Library (STL). gram to define a scanner and parser
expected delays, and the need to repair • Improvements in the management of that includes all elements of logical in-
newly discovered defects. GPSS imple- schedule network objects were made. ference (for example, AND, OR, and
ments constraint-based scheduling algo- An embedded schedule-data configu- NOT) as well as full capability for
rithms and provides an interactive ration-management subsystem, similar building and incorporating user-cus-
scheduling software environment. In re- to systems used for software configura- tomizable queries.
sponse to inputs, GPSS can respond with tion management, was built. This sub- • An improved report architecture was de-
schedules that are optimized in the system accommodates multiple ver- veloped. The prototype featured a sig-
sense that they contain minimal viola- sions and revisions of each schedule, nificant number of hard-coded user op-
tions of constraints while supporting the including direct descendants and tions, and too little care was taken
most effective and efficient utilization of branches. It also implements a concept initially to develop a consistent but flexi-
space-shuttle ground processing re- of user sessions that enables each user ble report architecture. The re-engineer-
sources. to maintain multiple current instances ing of the affected software components
The present version of GPSS is a prod- of the same schedule and full schedule involved design around a new report
uct of re-engineering of a prototype ver- data files with sizes of the order of class that contains attributes that de-
sion. While the prototype version 1MB. scribe the class of objects (e.g., tasks)
proved to be valuable and versatile as a • Improvements in calendar operations represented in a report, the presenta-
scheduling software tool during the first were made. The original implementa- tion style (e.g., Gantt chart or tabula-
five years, it was characterized by design tion required the full, time-series ex- tion), and the time frame of the report.
and algorithmic deficiencies that af- pansion of all calendars, giving rise to All report definitions are saved in files
fected schedule revisions, query capabil- a large memory overhead. Further- that the user can edit to customize re-
ity, task movement, report capability, more, some calendar features (e.g., ports.
and overall interface complexity. In ad- holidays), were “hard-coded.” In the • Several improvements in algorithms
dition, the lack of documentation gave re-engineering, calendar memory re- were made to solve backward-move-
rise to difficulties in maintenance and quirements were reduced by providing ment problems, provide a more robust
limited both enhanceability and porta- for all calendar calculations to be per- implementation of achievers, and im-
bility. formed in real time and by removing prove memory management through
The goal of the GPSS re-engineering all hard-coded elements. the use of smart pointers and “lazy
project was to upgrade the prototype • Re-engineering of a robust query sub- load” of persistent data. Also included
32 NASA Tech Briefs, January 2004
is an updated implementation of an creases portability. For the users, every ef- Lotti, James M. Moody, Tony K. Nguyen,
object-oriented callback system to the fort was made in the re-engineering to Kenneth A. Peterson, Susan Sargent, Karma
Motif widget set. maximize flexibility and improve upon Shaw, Mack D. Stoner, Deborah S. Stowell,
The benefits of the re-engineered ver- the intuitive nature of the interface with- Daniel A. Young, and James H. Tulley, Jr., of
sion of GPSS hinge on the object-ori- out sacrificing any of the capabilities that United Space Alliance for Kennedy Space
ented approach. The use of STL and the made the prototype successful. Center. For further information, contact the
improvements in schedule and query op- This work was done by Joseph A. Barretta, Kennedy Commercial Technology Office at
erations are incorporated in C++ libraries Earl P. Johnson, Rocky R. Bierman, Juan 321-867-8130.
that may prove useful on succeeding pro- Blanco, Kathleen Boaz, Lisa A. Stotz, KSC-12043
jects. The rewriting of software in C++ in- Michael Clark, George Lebovitz, Kenneth J.
Bayesian-Augmented Identification of Stars in a Narrow View
An adaptive threshold guides acceptance or rejection of a tentative identification.
NASA’s Jet Propulsion Laboratory, Pasadena, California
An algorithm for the identification of view only 2°wide. The present algorithm catalog to be the center star.
stars from a charge-coupled-device is based partly on one such prior algo- 2. Decide which star is the neighbor star.
(CCD) image of a star field has been ex- rithm, called the “grid algorithm,” that The neighbor star is deemed to be the
tended for use with narrower field-of- has shown promise for identifying stars star nearest to the center star outside a
view images. Previously, the algorithm in fields of view 8°wide. To make it pos- buffer radius of brpixels. The value of
had been shown to be effective at a field sible to identify stars in fields of view bris chosen on the basis of experience.
of view of 8°. This work augments the down to 2°with acceptably low probabil- 3. Center a grid of grows and gcolumns
earlier algorithm using Bayesian deci- ities of error, the grid algorithm has on the center star, and orient the grid
sion theory. The new algorithm is shown been extended by incorporating such that a horizontal vector from the
to be capable of effective star identifica- Bayesian decision theory. center to the right edge passes through
tion down to a field of view of 2°. The al- For the special purpose of the grid al- the neighbor star. Like br, the value of
gorithm was developed for use in esti- gorithm, the term “pattern” denotes a gis chosen on the basis of experience.
mating the attitude of a spacecraft and grid representation of the relative posi- 4. Derive a pattern, a g2-element bit vector
could be used on Earth to help in the tions of stars in a field of view. Each star
identification of stars and other celestial is deemed to be located within one of V[0. . . g2– 1]
objects for astronomical observations. the cells of a square grid that spans ei-
The present algorithm is one of sev- ther the field of view of the CCD image such that if grid cell(i,j) contains a star,
eral that seek matches between (1) im- or a candidate star-catalog field of the then
aged star fields and (2) portions of the same angular dimensions. The portion
sky, with angular dimensions equal to of the grid algorithm that generates a V[jg+ i] = 1
those of the imaged star fields, in a cata- pattern comprises the following steps
log of stars in a known reference frame. (see figure): The vector element corresponding to
Previously developed star-identification 1. Choose a star from the CCD image or any grid cell which does not contain a
algorithms are not suitable for fields of the applicable field of view in the star star is given the value 0. The dot product
Center grid on one star and orient grid
Mark each grid cell that contains a star.
with reference to nearest neighbor star.
Image or Star-Catalog Field Final Pattern
A Grid Patternis created from either a CCD image of stars or star-catalog data for a field of view of the same angular dimensions as those of the CCD image.
NASA Tech Briefs, January 2004 33