Integrated Analysis and Quality Framework for Automotive Embedded Software Quality

Heinz Hille

Doctoral thesis

Abstract:

Researchers and industry agree that using metrics is the best approach sofar for measuring and evaluating software quality. In fact, metrics, theirevaluation and validation are subject of interest of a number of researchers.Sneed et al. (2010) describe di fferent measurement scales and di fferentmetrics. Although a number of metrics have been investigated and validated,an observation of the author showed that many auditees feel inconvenientwith their audit result. And in fact, the result is not understood becausethe audit approach is adapted to quality management prerequisi tes for eachorganization and hence an individual concept is created.The following comment shows from an academic viewpoint the actuality ofmetrics and quality management: “Measurement, metrics, and statisticalanalysis of data are the basic tools of science and engineering. Unfortunately,the software industry has existed for more than 50 years with metricsthat have never been formally validated, and with statistical techniques thatare at best questionable.” (Jones 2006, p. 2) Even worse, according to Bieman(1996), Jones and Capers (2014), Florac (1992), Gopu(2003) the use ofsoftware metrics admits that currently used metrics are not in general reliable.Moreover , the researcher of the dissertation sees a need for a morecommon and plausible approach with metrics.Automotive industry needs new concepts for processing and working withmetrics. Metrics must be precise, based on theoretical knowledge. Furtherthe pedagogical influence on developers must be used to improve theirskills and let them commit on higher qual ity based on understandable metrics.This will result in a better use of metr ics in practice. For management,it important for the same reasons and addi tionally because of an aggregatednumber of metrics they can deal with in their non-technical function. Obviously,there is a lack of available and accessible information about detailson investigations made about metrics. In spite of that, metrics are in useand they make sense to evaluate and to validate software.Because of the mentioned weaknesses this dissertation proposes a di fferentapproach that is easier to understand. In that context, the fuzzy approach ischosen to comply wi th the needs of audi tees, auditors, and management.Fuzzy has a lot advantages, as the name ‘ fuzzy’ states in terms of demandedexactness. Further the calculation with less available measurements ispossible. From a pedagogical aspect, the use of fuzzy mathematics is easyfor explanation of results.All research results, approaches, processes, and methods are covered withina framework. This framework of a fuzzy metric system has more capabilitiesthan a simple system of metrics. Either a metrics system or even a metricframework must be explained to partners and suppliers. This is necessarybecause of the complex topics of technical issues in the indust ry today.However, the framework also explains the reasons for specific behaviorof metrics. Then i t helps to inspi re an inducement to organize developmentprocesses with respect to software quality topics. All these measuresbring a positive continuity into organizational development of quality management.

Abstract:

Researchers and industry agree that using metrics is the best approach sofar for measuring and evaluating software quality. In fact, metrics, theirevaluation and validation are subject of interest of a number of researchers.Sneed et al. (2010) describe di fferent measurement scales and di fferentmetrics. Although a number of metrics have been investigated and validated,an observation of the author showed that many auditees feel inconvenientwith their audit result. And in fact, the result is not understood becausethe audit approach is adapted to quality management prerequisi tes for eachorganization and hence an individual concept is created.The following comment shows from an academic viewpoint the actuality ofmetrics and quality management: “Measurement, metrics, and statisticalanalysis of data are the basic tools of science and engineering. Unfortunately,the software industry has existed for more than 50 years with metricsthat have never been formally validated, and with statistical techniques thatare at best questionable.” (Jones 2006, p. 2) Even worse, according to Bieman(1996), Jones and Capers (2014), Florac (1992), Gopu(2003) the use ofsoftware metrics admits that currently used metrics are not in general reliable.Moreover , the researcher of the dissertation sees a need for a morecommon and plausible approach with metrics.Automotive industry needs new concepts for processing and working withmetrics. Metrics must be precise, based on theoretical knowledge. Furtherthe pedagogical influence on developers must be used to improve theirskills and let them commit on higher qual ity based on understandable metrics.This will result in a better use of metr ics in practice. For management,it important for the same reasons and addi tionally because of an aggregatednumber of metrics they can deal with in their non-technical function. Obviously,there is a lack of available and accessible information about detailson investigations made about metrics. In spite of that, metrics are in useand they make sense to evaluate and to validate software.Because of the mentioned weaknesses this dissertation proposes a di fferentapproach that is easier to understand. In that context, the fuzzy approach ischosen to comply wi th the needs of audi tees, auditors, and management.Fuzzy has a lot advantages, as the name ‘ fuzzy’ states in terms of demandedexactness. Further the calculation with less available measurements ispossible. From a pedagogical aspect, the use of fuzzy mathematics is easyfor explanation of results.All research results, approaches, processes, and methods are covered withina framework. This framework of a fuzzy metric system has more capabilitiesthan a simple system of metrics. Either a metrics system or even a metricframework must be explained to partners and suppliers. This is necessarybecause of the complex topics of technical issues in the indust ry today.However, the framework also explains the reasons for specific behaviorof metrics. Then i t helps to inspi re an inducement to organize developmentprocesses with respect to software quality topics. All these measuresbring a positive continuity into organizational development of quality management.

Keywords

Metriky, Měření, Management kvality

Keywords

Measurement, Metrics, Management

Language used: English

Date on which the thesis was submitted / produced: 8. 1. 2018

Thesis defence

• Supervisor: doc. Ing. Pavel Žufan, Ph.D.
• Reader: Tomáš Macák, doc. Ing., Ph.D., Radim Farana, prof. Ing., Milan Fekete, externi

Mendel University in Brno

Faculty of Business and Economics

Doctoral programme / field:
Economics and Management / Business Management and Economics