Chapter 5 Robust Performance Tailoring with Tuning - SSL - MIT
Chapter 5 Robust Performance Tailoring with Tuning - SSL - MIT
Chapter 5 Robust Performance Tailoring with Tuning - SSL - MIT
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
a statistical robustness measure such as standard deviation into the objective. The<br />
RPT optimizations are applied to the SCI development model, and it is found that<br />
anti-optimization provides the most robust design when the uncertainty is modeled<br />
as a bounded uniform distribution.<br />
The resulting RPT design is compared to the PT design and it is found that the<br />
worst-case performance of the RPT design is significantly lower than that of the PT<br />
design. However, the increase in robustness comes at the cost of nominal performance,<br />
so that the RPT has a higher RMS performance at the nominal uncertainty values<br />
than the PT design. A comparison of the worst-case performance values of the two<br />
designs over a range of uncertainty bounds shows that RPT designs can tolerate<br />
greater variation in the uncertainty parameters at a specific performance requirement<br />
than the PT design. However, since nominal performance is sacrificed, RPT designs<br />
cannot meet aggressive performance requirements given high levels of uncertainty.<br />
The RPT designs fail when uncertainty is high because it is not possible to find val-<br />
ues of the tailoring parameters that meet requirements under all possible uncertainty<br />
realizations. Therefore, in <strong>Chapter</strong> 4, the concept of dynamic tuning is introduced.<br />
It is defined as the adjustment of physical tuning parameters on hardware that brings<br />
its performance to <strong>with</strong>in required values. A tuning optimization is formulated and<br />
applied to the PT and RPT designs in the worst-case uncertainty realizations. It<br />
is shown that in both cases, dynamic tuning succeeds in reducing the performance<br />
variance of the worst-case design, however, the effect is more dramatic on the PT<br />
design. The worst-case uncertainty realizations of both designs, over a range of un-<br />
certainty bounds, are tuned, and it is shown that dynamic tuning extends the range<br />
of performance requirements that each design can meet at a given uncertainty level.<br />
The practical issues associated <strong>with</strong> dynamic tuning are also considered. In reality<br />
only limited performance data is available from the hardware and the actual values<br />
of the uncertainty parameters are not known. Therefore some type of model up-<br />
dating technique or real-time hardware tuning is necessary to improve the hardware<br />
performance. A set of possible hardware tuning methods, ranging from model-only<br />
methods that rely on the uncertain model to find successful tuning configurations to<br />
213