Software Engineering

183
10.4 
Documenting the functional architecture
to force the selection of a desired sequence, or probabilistic to randomly select a 
sequence to execute. The behavioral model may include a probability distribution 
equation to represent the random selection. The following is a list of common prob-
ability distribution functions:
●
  Bernoulli distribution
●
  Binomial distribution
●
  Uniform (discrete) distribution
●
  Poisson binomial distribution
●
  Geometric distribution
●
  Logarithmic distribution
●
  Exponential distribution
●
  Pareto distribution
●
  Chi-square distribution
●
  Weibull distribution
An executable behavior model can be utilized to support design trade-off 
analysis, an element of the software analysis practice (see Chapter 14). Rather than 
relying on abstract diagrams, an executable model provides a working prototype of 
the software product. Competing design alternatives can be compared to determine 
which design approach performs best under anticipated operational conditions. A 
Monte Carlo simulation can be achieved by enclosing the model in a looping con-
trol structure and conducting repeated analysis of functional timing and resource 
utilization. This enables the probabilistic identification of critical paths through the 
model that can be used to guide the development of software test cases. The execut-
able model also is valuable in verifying that the functional architecture satisfies the 
specified software requirements.