A Technical History of the SEI
ihQTwP
ihQTwP
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
analysis as well as model tuning for runtime deployment <strong>of</strong> Q-RAM models. Guidance was developed<br />
for model tuning <strong>of</strong> <strong>the</strong> Q-RAM. Model tuning involves specifying a consistent set <strong>of</strong> Q-<br />
RAM model parameters, including <strong>the</strong> utility values for quality dimensions and task weights. A<br />
notation was also developed to support <strong>the</strong> expression <strong>of</strong> cross-task quality-level constraints to<br />
achieve predictable degradation behavior [Rajkumar 1997]. The Visual Q-RAM s<strong>of</strong>tware tool was<br />
enhanced to support use-scenario walkthroughs and development<br />
<strong>of</strong> models from predefined libraries <strong>of</strong> task<br />
types. Use <strong>of</strong> this technology in DoD contexts was<br />
demonstrated by developing a set <strong>of</strong> examples; <strong>the</strong>se included<br />
helicopter pilot mission support, phased array radar<br />
bandwidth allocation, and radar target-tracking algorithm<br />
selection.<br />
Real-time queueing <strong>the</strong>ory provides accurate timing behavior<br />
predictions <strong>of</strong> real-time systems having stochastic<br />
workloads. The <strong>the</strong>ory can be used to assess <strong>the</strong> ability<br />
<strong>of</strong> a system to meet <strong>the</strong> timing requirements under heavy<br />
traffic conditions. It complements scheduling <strong>the</strong>ories<br />
such as generalized rate monotonic scheduling. Visual<br />
RTQT is a tool that demonstrates <strong>the</strong> practicality <strong>of</strong> using<br />
RTQT. Important progress was made in <strong>the</strong> development<br />
<strong>of</strong> RTQT and its application to avionics and communication<br />
systems [Lehoczky 1996]. RTQT is a<br />
significant innovation in <strong>the</strong> design and scheduling <strong>of</strong><br />
real-time systems in that it is capable <strong>of</strong> making exact<br />
predictions <strong>of</strong> a system’s ability to meet <strong>the</strong> timing requirements<br />
<strong>of</strong> real-time tasks where task arrivals and<br />
computation requirements are stochastic. It extends<br />
methodologies such as GRMA to a greatly broadened<br />
framework. The RTQT project innovations include development<br />
<strong>of</strong> an RTQT-based analysis <strong>of</strong> <strong>the</strong> temporal<br />
behavior <strong>of</strong> <strong>the</strong> F-22 avionics challenge problem and an<br />
RTQT analysis for feed-forward queuing networks and<br />
for acyclic networks, leading to an analysis tool for <strong>the</strong><br />
determination <strong>of</strong> end-to-end schedulability requirements.<br />
The View from O<strong>the</strong>rs<br />
This group [CMU IMPACT]<br />
demonstrated all proposed objectives<br />
from <strong>the</strong> DASADA literature.<br />
This group works closely with<br />
Lockheed Martin on real time<br />
scheduling and context testing on<br />
<strong>the</strong> F-16 avionics platforms. CMU<br />
is doing breadboard testing and<br />
creating prototypes for a new advanced<br />
avionics suite proposed for<br />
future aircraft development. An<br />
evaluation <strong>of</strong> this system indicated<br />
this group is ready to move on to<br />
<strong>the</strong> next phase <strong>of</strong> <strong>the</strong> DASADA<br />
program. (pp. 63-64)<br />
Out <strong>of</strong> <strong>the</strong> 19 projects, <strong>the</strong>re is<br />
only a handful that should be considered<br />
for future funding based<br />
upon <strong>the</strong>ir level <strong>of</strong> effort over <strong>the</strong><br />
past several months, as well as<br />
<strong>the</strong>ir level <strong>of</strong> technology maturity<br />
to be able in <strong>the</strong> next year to actually<br />
provide a component to insert<br />
into <strong>the</strong> DASADA Dynamic Assembly<br />
Toolkit. [One <strong>of</strong> those projects<br />
is] CMU’s Integrated Methods for<br />
Predictive Analytic Composition<br />
and Trade<strong>of</strong>f (IMPACT). (pp. 76-<br />
77)<br />
– Wayne S. Mandak and<br />
Charles A. Stowell,<br />
[Mandak 2001]<br />
Hierarchical scheduling is a method based on GRMA<br />
that enables a single schedulable physical resource, such<br />
as a processor, to be partitioned into multiple isolated<br />
virtual resources. Different algorithms and analysis techniques<br />
can be used in each <strong>of</strong> <strong>the</strong> virtual resources, and<br />
changes in <strong>the</strong> temporal properties within one virtual resource<br />
do not impact <strong>the</strong> temporal behavior within o<strong>the</strong>r virtual resources [Saewong 2002].<br />
CMU/<strong>SEI</strong>-2016-SR-027 | SOFTWARE ENGINEERING INSTITUTE | CARNEGIE MELLON UNIVERSITY 51<br />
Distribution Statement A: Approved for Public Release; Distribution is Unlimited