SOFTWARE DESIGN FOR X - Software Design for Six-Sigma: A Roadmap for Excellence - page 365

TABLE 14.2 Software Reliability Growth Models

Formula for
Hazard Function

Data and/or Estimation
Required

Limitations and
Constraints Model Name

Number of detected
faults at some time x
( µ ).
Estimate of λ 0

Software must be
operational.
Assumes no new
faults are introduced
in correction.
Assumes number of
residual faults
decreases linearly
across time.

Musa Basic λ 0 [1 − µ / ν 0 ]

Number of detected
faults at some time x
( µ ).
Estimate of λ 0
Relative change of
failure rate over time
( φ )

Software must be
operational.
Assumes no new
faults are introduced
in correction.
Assumes number of
residual faults
decreases
exponentially across
time.

Musa Logarithmic λ 0 exp( − φµ )

Number of corrected
faults at some time x.
Estimate of E 0

Software must be
operational.
Assumes no new
faults are introduced
in correction.
Assumes number of
residual faults
decreases linearly
across time.

General
Exponential
(General form of
the Shooman,
Jelinski-
Moranda, and
Keene-Cole
exponential
models)

K(E 0 − E c (x))

α

Estimate of α (number
of failures)
Estimate of
(reliability growth)
Time between failures
detected or the time of
the failure occurrence.

Software must be
operational.
Assumes uncertainty
in correction process.

Littlewood/Verrall

( t + ( i ))

Faults detected in
equal interval i

Software must be
operational.

Schneidewind
model

α exp ( − βι )

( Continued)