Page 159 - DCAP305_PRINCIPLES_OF_SOFTWARE_ENGINEERING

Unit 7: Introduction to Verification

Self Assessment Questions Notes
1. The terms verification and validation have different meaning depending on the discipline
in engineering or quality management systems.
( a) True (b) False
2. ...................... are established during the requirements phase of the conceptual model
development and incorporate numerical and experimental uncertainty.
( a) verification metric (b) validation metric
( c) verification and validation metric (d) None of these
3. ......................... are established during the requirements phase of the conceptual model
development and incorporate numerical and experimental uncertainty.
( a) Network metrics (b) Validation metric
( c) Verification and validation metric (d) Verification metric
4. The Verification and Validation model commonly known as V Model is considered to be
an extension of the.......................
( a) phenomenological models (b) domain model
( c) computational models (d) Waterfall model
5. The biggest disadvantage of V-model is that it is not very rigid and the least flexible.
( a) True (b) False

7.2 Meaning of Metrics

Metric of attention is complexity, as one of our goals is to strive for straightforwardness and
ease of understanding. A possible use of complexity metrics at design time is to improve the
design by plummeting the complexity of the modules that have been established to be most
complex. This will directly improve the testability and maintainability. If the complexity cannot
be reduced because it is inherent in the problem, complexity metrics can be used to highlight
the more complex modules. As complex modules are often more error-prone, this feedback can
be used by project management to ensure that strict quality assurance is performed on these
modules as they evolve. Overall, complexity metrics are of great interest at design time and
they can be used to evaluate the quality of design, improve the design, and improve quality
assurance of the project. We will describe some of the metrics that have been proposed to
quantify the complexity of design.
A Validation metric is the basis for comparing features from experimental data with model
predictions. Validation metrics are established during the requirements phase of the conceptual
model development and incorporate numerical and experimental uncertainty. If the error, e ,
between experimental data, y , and model prediction, *y , is given by *e = y − y , a simple metric
could be the expected value of the error, E( ) e , or the variance of the error, V( ) e . Other metrics
could include, for example: P ( ) e > 0 , where P( ) is the probability.
Percentiles on the probability distribution of e; or a hypothesis test such as E ( ) e > 0, where
the validation metric is a pass/fail decision of whether or not the model is contradicted by the
data. In selecting the validation metric, the primary consideration should be what the model
must predict in conjunction with what types of data could be available from the experiment.
Additionally, the metrics should provide a measure of agreement that includes uncertainty
requirements. Complex model simulations generate an enormous amount of information from
which to choose. The selection of the simulation outcome should first be driven by application
requirements. For example, if a design requirement is that the peak strain at specified location
should not exceed some value, and then the model validation should focus on comparison
of measured and computed strains at that location. Features of experimental data and model
outputs must be carefully selected. A feature may be simple, such as the maximum response

LOVELY PROFESSIONAL UNIVERSITY 153