Other significant
building blocks for the model are the environmental and protocol parameters
that have an significant impact on performance. As we are aiming
to build an abstracted, simplified model of reality, this need not be an
exhaustive list, but should include key parameters. This is more of an art
than a science in many respects, and often entails an iterative process to
refine the components of the model.
4. Derive a simple mathematical expression that gives the performance metric
for that protocol, as a function of the variables corresponding to key
environmental, network, and protocol parameters. Along with the previous
step, this is part of the core work of building the model.
5. Refine the model, by adding, discarding, or modifying variables corresponding
to environmental, protocol or network settings. The goal is to
obtain an expression for the protocol performance metric that illustrates
the core tradeo?®.
6. Solve for the value of the protocol parameter which optimizes the performance
metric of interest. There are often opposite trends in di?®erent
components of the model that are in tension with each other. As a result,
performance loss may be incurred in setting a key protocol parameter to
too low a value or too high a value. In most cases, determining the optimum
parameter setting requires just finding the zero of the derivative
of the metric with respect to the parameter in question.
Pages:
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618