emery@ARIES.MITRE.ORG (David Emery) (02/15/89)
Reprinted with permission from Federal Computer Week (Feb 13, 1989),
copyright (c) 1989 by IDG Communcations Inc. All rights Reserved.
(thanks to FCW for permission to reproduce this here. typos are mine
dave emery)
---------------
Army's AFATDS Passes Formal Qualifications Test
by Fred Reed
The Army's Advanced Field Artillery Tactical Data System, for years
the object of controversy, has passed its formal qualification test
without any major problems, according to both the Army and Magnavox,
developer of the system.
AFATDS, chiefly a software product, will run mostly on standard
Army field computers but with some special-purpose hardware. Now
running on test-bed equipment from Magnavox, AFATDS automates fire
control of artillery. The system receives requests for artillery fire
from units in the field, decides which are most important in
accordance with the battery commander's instructions, and directs fire
to the target.
AFATDS, a 1.2 million-line Ada program, has been criticized
heavily for software problems. Predictions have been made that it was
too ambitious and would never work.
However, in extensive testing, according to Lt. Col. David Peterson
of the Army's tactical-computer establishment in the Pentagon, "It did
well. We found some things to fix but nothing remotely catastrophic."
Specific Ratings
In the test, involving five artillery scenarios running a week
each, during which 854 functions were tested, the software suffered 16
priority-three software trouble reports, and 15 priority-four STRs,
said John Williams, program manager for AFATDS at Magnavox.
He said a priority-one STR is serious enough that the software
cannot pass under any circumstances. A priority-two STR prevents
passing unless it can be explained. AFATDS had no ones or twos. The
other classes of STR are of diminishing seriousness. The defects
added to 14 percent of the maximum allowed.
Ada, the mandated programming language for defense use, got high
marks from Magnavox. In the past Ada had been criticized as too bulky
and slow for real-time use, and AFATDS was considered too complex to
be practical.
"My software director is adamant that if it had not been for the
power of Ada, we never could have completed the program. We were
particularly pleased with how short a time it took to find errors.
The language and development tools are getting better. So are
compilers, " Williams said.
Independent Evaluation
The Army's Communications Electronics Command had AFATDS rated by
an independent evaluator, Dynamics Research Corp., to see how it
compared with industry standards, Williams said. AFATDS rated as
follows on different measures: Error handling, average; modularity,
excellent; clarity, good to excellent; self-descriptiveness, good to
excellent; independence, excellent; simplicity, very good, he said.
Ralph Crafts, editor of the "Ada Strategies" newsletter and a
strong advocate of Ada, said "Magnavox did everything right
technically. The criticism was political meddling by Litton
[Industries, Inc., the developer of a competing system]."
However, Army officials said serious management problems plagued
Magnavox early in the program. Some said AFATDS was close to being
excessively ambitious. One officer said, "For a while we thought we
had asked too much, so it's good to see that it looks doable after
all." The next step, Peterson said, will be concept evaluation, with
fielding expected in the mid-1990s.