Evaluation of standard reliability growth models in the context of automotive software systems
Presented at:
PROFES conferences, the 14th International Conference of Product Focused Software Development and Process Improvement, in Paphos, Cyprus, 12-14 June 2013.
Introducing Microsoft’s new Enterprise Work Management (EWM) Solution
Evaluating SRGMs for Automotive Software Project
1. Evaluation of standard reliability growth models in
the context of automotive software systems
SRGMs: Software Reliability
Growth Models
Rakesh Rana1, Miroslaw Staron1, Niklas Mellegård1, Christian Berger1,
Jörgen Hansson1, Martin Nilsson2, Fredrik Törner2
1Software Engineering division,
Department of Computer Science and Engineering,
Chalmers/ University of Gothenburg
2Volvo Cars Corporation
2. This Car Runs on Code
“It takes dozens of mircroprocessors running 100 million lines of
code to get a premium car out of the driveway, and this software is
only going to get more complex” -ieee spectrum
Ref: http://spectrum.ieee.org/green-tech/advanced-cars/this-car-runs-on-code
3. Reliability
*Reliability and dependability are very important features
of any computer system.
*Have we done enough testing?
*Is the software ready for release?
*How should we adjust/optimize our testing strategy?
SRGM -> Software Reliability and Maturity
SRGM -> Use for Automotive Software Projects
4. Data used (Automotive Project)
Mellegård, N., Staron, M., and Törner, F.: ‘A light-weight defect classification scheme for embedded
automotive software and its initial evaluation’
5. Different Software Reliability Growth Models
Model Name Model Type Mean Value Function Reference
Models with 2 parameters
Goel-Okumoto (GO) Concave 𝑚 𝑡 = 𝑎(1 − 𝑒−𝑏𝑡
) [11]
Delayed S-shaped model S-shaped 𝑚 𝑡 = 𝑎(1 − (1 + 𝑏𝑡)𝑒−𝑏𝑡
) [12]
Rayleigh model 𝑚 𝑡 = 𝑎𝑒−𝑏/𝑡
Models with 3 parameters
Inflection S-shaped model S-shaped
𝑚 𝑡 =
𝑎(1 − 𝑒−𝑏𝑡
)
(1 + 𝛽𝑒−𝑏𝑡 )
[9]
Yamada exponential imperfect
debugging model (Y-ExpI)
S-shaped
𝑚 𝑡 =
𝑎𝑏
∝ + 𝑏
(𝑒∝𝑡
− 𝑒−𝑏𝑡
)
[13]
Yamada linear imperfect
debugging model (Y-LinI)
S-shaped 𝑚 𝑡 = 𝑎 1 − 𝑒−𝑏𝑡
1 −
∝
𝑏
+ ∝ 𝑎𝑡 [13]
Logistic population model S-shaped 𝑚 𝑡 =
𝑎
1 + 𝑒−𝑏 𝑡−𝑐
[14]
Gompertz model S-shaped 𝑚 𝑡 = 𝑎𝑒−𝑏𝑒−𝑐𝑡
[15]
12. Conclusions and further work
*Two parameters models: fit - reasonable, asymptotes -
unrealistic;
*Logistic and inflectionS: Best fit to our data among the
different models tried;
*Important factors: Using appropriate time scale.;
*Using parameter estimates from two parameter models
and current project information, can give useful insight for
optimizing the resource allocation going forward.
13. Summary and Impact
*Logistic and inflectionS and Gompertz model gives best
fit and asymptote predictions.
*Identifying right models and using SRGMs in the
company and automotive sector in general will:-
*Help assess the reliability of software developed and thus the
release readiness.
*Using SRGM during the project can help test and quality
managers to make optimal testing resource allocation decisions.
*Thus correct use of SRGMs help the company & the automotive
industry to develop and release high quality software.