Abstract
Title: " The Use of Mosaic as a Documentation Tool For Large,
Graphically-Based Simulation Software"
Author: Donald J. Fabozzi II
Rome Laboratory, Advanced Radar Systems Branch
Large software tools, such as Rome Laboratory's "Space-Time
Adaptive Processing Algorithm Development Tool" (RLSTAP/ADT) require
Interface Control Documents (ICDs) to define and control the underlying
code and the user interfaces. The RLSTAP/ADT is typical of many large
software simulation tools in that it contains a large library of
routines, was developed at more than one site, and is highly graphical.
Subsequently, the ICD for programs such as these becomes very lengthy,
difficult to peruse, and requires special tools to edit and display the
code parameters with the graphical information. Editors which can
display both textual and graphical information are both expensive and
host specific. As a result, the ICD, which serves as the tool's
structural backbone, can become out of sync with tool development
versions or with ICD versions at other sites. Public domain tools like
Mosaic solve these problems through advanced editors and interfaces.
Along with providing remote access, Mosaic displays and manages the ICD
more effectively and more closely binds the development tool with the
ICD. This paper discusses how public domain tools like Mosaic improve
the representation and management of large software simulation tools
like the RLSTAP/ADT through an advanced display tool. This paper will
also discuss derived advantages of this approach such as improved
simulation development, flexibility, and publicity. In hopes of
promoting a standarized documentation environment, this paper will also
measure the effectiveness of the Mosaic environment against the
migration cost.
I. Introduction
Declining Government spending has prompted software simulations
as valuable design tools now more than ever before. Rome Laboratory(RL)
leads the development of simulations in the Command, Control,
Communications and Intelligence (C3I) programs as well as numerous radar
simulations. The Surveillance Directorate of RL recently converted one
of its very large radar simulation tools into a graphical, user-friendly
environment. This tool, the Rome Laboratory Space-Time Adaptive
Processing/ Algorithm Development Tool (RLSTAP/ADT), like many
simulations, has an Interface Control Document (ICD) which describes the
module interfaces.
The ICD is the backbone for a simulation in that it defines the
structure and module interfaces of the tool. One of its major functions
is to serves as a guide for the software development team by providing
module descriptions, i/o parameters, and pseudocode. During the
development of the RLSTAP/ADT, however, the ICD became difficult to
manage and update through available editors when the simulation became
very large and graphical. The simulation team surveyed commercial
products to manage this documentation at a reasonable price and found
nothing. The team discovered World Wide Web (WWW) tools and Mosaic to
be the best match for editing, browsing, and managing the simulation's
documentation.
This paper describes how converging the ICD to a World Wide Web
resource significantly improves the representation of the RLSTAP/ADT.
This paper also looks at the conversion cost and process and stresses
that this is an excellent documentation tool for all simulations of this
type.
II. Background
The RLSTAP/ADT is a total radar simulation
environment, consisting of radar data, signal processing routines, and
a flexible graphical interface. The routines, which fall within ten
major categories and about twenty or so sub-categories, are organized in
the ICD in a hierarchical fashion through "dotted" numerals. Below
displays a section from the ICD table of contents:
3.1 Physical Model
3.1.1 PMEnv - Physical Model
Environment Setup
3.1.2 RCVR_SET - Receiver Setup
3.1.2.1 RcvPlat - Receive
Platform Setup
3.1.2.2 RcvAnt - Receive Antenna Pattern
3.1.3
XMTR - Transmitter
3.1.3.1 XmtPlat - Transmit Platform Location and
Orientation
3.1.3.2 XmtAnt - Transmit Antenna Pattern
3.1.3.3
XmtWave - Transmit Waveform
The ICD is very hierarchical but too lengthy to include in
this report. However, the following graphic illustrates the
hierarchical depth and breadth of the ICD through one instance, the
Physical Model-Receiver module.
The ICD contains six sections of description for each module at
the lowest level of this hierarchy. These consist of a text
description, control parameters, input parameters, output parameters,
error checking, and pseudocode. Because the number of modules grew to
over 200, management of 1200 destinct sections of a document through a
generic text editor became difficult. The problem was compounded when
trying to match the graphical interface with the routine.
In addition, software reviews were awkward because hard copies of
the ICD resulted in numerous pages of text description, followed by
several pages of hand-drawn graphics.
III. The Conversion
Mosaic and HyperText Markup Language (HTML) met the project`s
needs in cost and functionality. Site license fee was an issue to the
RLSTAP/ADT team because the tool was developed at three geographically
separated locations with more participants expected to contribute to
the tool. HTML handles hierarchical structure very effectively through
hyperlinks so the user can easily browse through the depths of this
document.
Having the simulation's graphic interfaces displayed alongside
the corresponding module descriptions enhances the module description
considerably. Below is a page in the ICD which describes the main
section the Physical Model and displays its corresponding pull-down menu
from the graphical interface:
PM - Physical Model
The PM functional group contains modules that allow the user to
specify an operational environment for the scenario. The physical model
consists of ten items, each represented by one or more icons labeled
with an abbreviated name. The items in this functional group are: Setup,
Environment, Receive, Transmit, Atmospheric, Clutter, Target, Jammer and
Summer and Antenna Pattern. The pull-down window from the main
RLSTAP/ADT workspace is shown below:
In addition, Mosaic allows the placing of the module graphical
interface alongside the text description. As shown, one can quickly and
easily review both the module information and graphical interface:
3.1.2.1:Rcv_Plat - Receive Platform Setup
In comparison with the benefits derived from the conversion to
HTML, the conversion cost was minimal. Since Mosaic and the editors are
freely distributed, the cost equated the time spent in the conversion.
Contrary to popular opinion of public domain software, the time spent on
the conversion was minimal. Minus the time used in acquiring the public
domain tools and understanding the HyperText Markup Language, the
conversion took only about three days. This is due to the fact that the
ICD was already sectioned numerically and the conversion could be
automated through a unix shell script.
This process proved far more cost advantageous than using
commercial software or building a custom product. Both were expensive
in that commercial graphical packages sell for $500 to $2000 per
license, and custom software would have easily cost $10,000. It is also
important to note that learning costs would have been incurred in
addition to these prices regardless of the quality of the commercial or
custom product.
IV. Conclusions
This paper discusses and displays the
advantages of converting a simulation tool's Interface Control Document
into HyperText Markup Language for viewing through WWW browsers such as
Mosaic. HTML and Mosaic dramatically improved this project's
documentation by handling the hierarchical and graphical nature of the
document. In addition, since the RLSTAP/ADT will soon be publicly
distributed, Mosaic will facilitate promotion, background information
and tutorials.
This paper was motivated by this project's accomplishments and
also to promote this documentation process to the simulation community.
Author information
Donald J. Fabozzi II (D.J.) works in the
Advanced Radar Systems Branch of Rome Laboratory. D.J. is the chief
software engineer behind the Radar simulation tool "Rome Laboratory
Space Time Adaptive Processing /Algorithm Development Tool". He earned
a B.A. degree in Mathematics '87 and a B.S. degree in Electrical
Engineering '89 at Potsdam College and Clarkson University,
respectively. D.J. went on to receive his M.S. in Computer Engineering
'93 from Syracuse University. While at Rome Laboratory, he performs a
variety of tasks, including radar analysis coding, system design, and
system administration. D.J.'s expertize includes Unix, C, Khoros and
networking.
D.J. can be reached at fabozzid@lonexb.rl.af.mil