Towards Reducing the Complexity of Adaptive Real-Time
Large-Scale Distributed Embedded Systems
Citation: Lisa DiPippo, Jiangyin Zhang, Matthew Murphy, Victor Fay Wolfe, Joseph Loyall, Richard Schantz, Craig Rodrigues, Jeff Parsons, Sandeep Neema, Balachandran Natarajan, and Aniruddha Gokhale. IEEE Workshop on Large Scale Real-Time and Embedded Systems, in conjunction with IEEE Real-Time Systems Symposium, December, 2002, Austin, Texas.
Formats: Word
Abstract
This
paper describes elements of the approach that we are taking to address the
complexity inherent in creating software for large scale distributed real-time
embedded (LDRE) applications such as the control of total ship computing on the
new US Navy surface ships (DDX), coordinated unmanned vehicles,
meteorological measurement and
prediction systems, and widely distributed automated financial control. These applications are functionally complex,
and their complexity is further amplified due to non-functional considerations
such as:
-
their
large scale
-
their
potentially wide distribution
-
their
real-time requirements
-
their
dynamically changing environment
In
general, our approach is to provide more of the solution to the non-functional
aspects commonly available off-the-shelf through a combination of advanced
middleware services and advanced software engineering approaches, which when
combined would provide for developing these systems from a much higher level
(and presumably less costly and error prone) basis. We briefly discuss how we have begun to apply four inter-related
concepts to the coordinated, real-time scheduling component of our larger QoS
management approach. The four concepts are:
-
The
use of standards-based middleware
-
Building
in run-time adaptability into scheduling
-
“Weaving”
scheduling into applications using aspect-oriented programming
-
A
model-integrated computing approach to scheduling
