Observational Models of Requirements Evolution CSC2004 Light

8/3/2019 Observational Models of Requirements Evolution CSC2004 Light

1/36

Massimo Felici

2004

CSC2004

Observational Models of

Requirements Evolution-

Theory and Practice

Massimo FeliciLFCS, School of Informatics, The University of Edinburgh

[email protected]

http://homepages.inf.ed.ac.uk/mfelici/

The Critical System Conference

29th-30th September 2004


2/36

Massimo Felici2004

CSC2004

Overview

o On Requirements Evolutiono Socio-technical grounds

o Empirical analyseso Capturing Requirements Evolution

o Observational Models of Requirements Evolutiono A bit of theory

o Evolution as Dependability


3/36

Massimo Felici

2004

CSC2004

On Requirements Evolution

Requirements Evolution is one of the main issues thataffect development activities as well as systemfeatures (e.g., system dependability). Althoughresearchers and practitioners recognise theimportance of requirements evolution, research resultsand experience are still patchy. This points out a lackof methodologies that address requirements evolution.


4/36

Massimo Felici2004

CSC2004

On Requirements Evolution

o Do requirements evolve? Yes!!!

oCan requirements be frozen? No

o you might try, but it wouldnt work

o The completeness and correctness panaceas are just utopias

o Why do requirements evolve?o It is possible further to understand requirements evolution from a

socio-technical viewpointo Systems Approach, Heterogeneous Engineering, Social Shaping of

Technology, etc.

o Empirical analyses of requirements evolution are still patchyo Lack of support for analyzing evolutionary data

o Organizational culture/strategyo Requirements evolution (analysis) as work practice

o Requirements evolution captures organizational knowledge


5/36

Massimo Felici

2004

CSC2004

Capturing RequirementsEvolution

Everyone talksabout Requirements

Evolution, but justfew know how itlooks like.

Everyone talksabout Requirements

Evolution, but no

one knows how itlooks like.

No one talks aboutRequirements

Evolution.


6/36

Massimo Felici2004

CSC2004

Capturing Requirements Evolution [1/10]

o Requirements evolution as trends of requirements changeso added, deleted and modified requirements

o An increasing number of requirements is due too Stakeholders interaction, technology issues/constraints,

further understanding of system requirements,Requirements refinements, new arising requirements,Requirements dependencies, operational usage, etc.


7/36

Massimo Felici2004

CSC2004

o Requirements (changes) classification

o A classification is a spatial, temporal or spatio-temporalsegmentation of the world. In an abstract, ideal sense, aclassification system exhibits the following properties:There are consistent, unique classificatory principles inoperation; The categories are mutually exclusive; Thesystem is complete.

[Bowker and Star, 1999]

o Classifications and Standards are closely related, but notidentical.o A standard:

o agreed-upon rules, multi-community (users) of practice;distant and heterogeneous collaborations; legal bodies,lack of laws (or criteria) of definition (e.g., politics)

o Classifications:o capture work practice; are ubiquitous; represent information

infrastructures; etc.; evolve.



8/36

Massimo Felici2004

CSC2004


o Example of Requirements (changes) classificationo Type of Requirement:

o Stable or Changing (Mutable, Emergent, Consequential,Adaptive, Migration)

o Type of change:

o Add, Delete Modify; Partial compliance; Traceability; etc.

o Change Rationale:

o requirements errors, conflicts and inconsistency; evolvingcustomer system knowledge; technical, schedule or cost

issues; changing customer priorities; environmentalchanges; organizational changes


9/36

Massimo Felici2004

CSC2004


o Requirements (changes) classifications allow the

analysis of requirements features, in particular,requirements evolution

Classified RequirementsChanges per Software

Release

Histograms of the size of the sets ofrequirements changes allocated to a single

software release


10/36

Massimo Felici2004

CSC2004


o Requirements (changes) classifications allow theanalysis of requirements features, in particular,requirements evolution

R

CR

R

RMI

=

HRMI - Historical RequirementsMaturity Index

R

CAR

R

HRMI

=

RMI RequirementsMaturity Index


11/36

Massimo Felici2004

CSC2004

Capturing Requirements Evolution [6/10]o Stable and Changing Requirements A functional viewpoint

points out how diverse parts of requirements evolve. Moreover,

it supports the analysis of requirements dependencies.

(Functional) Requirements size

A Functional Viewpoint


12/36

Massimo Felici2004

CSC2004

Capturing Requirements Evolution [7/10]o Requirements Evolution Processes A functional viewpoint

points out how diverse parts of requirements evolve. Moreover,

it supports the analysis of requirements dependencies.

Requirements dependencies

Requirements evolution processes

gamma maps from sequence analysis


13/36

Massimo Felici2004

CSC2004


o It is actually possible to identifytwo main distinct processes:

o the software process

o the requirements process

[Weinberg, 1997]

o Empirical evidence should mutually control the two processes

o A product viewpoint identifying reusable software functionso Repeatable processes to allocate software functions to

high level system requirements

o Refining requirements information flow expansion

o The gap between these processes represents the extent to whichan organization is able to identify an optimal and effective set of(reusable) software functions

o Issues: communications across multilevel organizations, cost,visibility, controllability, etc.


14/36

Massimo Felici2004

CSC2004


o The Humanso Viewpoints Analysis

different professionalprofiles (in terms ofresponsibility andexperience) point outdifferent understandings

o Viewpoint Analysis -Conflicting viewpointshighlight issues inrequirements process


15/36

Massimo Felici2004

CSC2004


o Requirements Evolution Practice

o Data Collection: difficult; integrated in work practice;otherwise, issues - increased workload, frustration, etc.

o Tailored data organizations and goals

o Requirements Evolution Featureso Quantitative Requirements Evolution

o Requirements maturity (e.g., RMI, HRMI, etc.)

o Taxonomy of requirements (changes)

o Functional requirements evolution

o Requirements evolution processes

o Requirements evolution viewpoints

o Require viewpoint management support

o Identify requirements issues

o Requirements evolution affects project visibility


16/36

Massimo Felici

2004

CSC2004

Modeling RequirementsEvolution

Heterogeneous Engineering: People had to be engineered, too persuaded to suspend their doubts, induced to provide

resources, trained and motivated to play their parts in aproduction process unprecedented in its demand. Successfullyinventing the technology, turned out to be heterogeneousengineering, the engineering of social as well as the physicalworld.

[MacKenzie, 1990]


17/36

Massimo Felici2004

CSC2004

Modeling Requirements Evolution [1/8]

o Heterogeneous Engineeringo Provides a comprehensive account of system requirementso Stresses a holistic viewpoint that allows us to understand

the underlying mechanisms of evolution of socio-technicalsystems

o Requirements, as mappings between socio-technical solutionsand problems, represent an account of the history of socio-technical issues arising and being solved within industrialssettings

o Requirements are socially shaped (I.e., constructed andnegotiated) through sequences of mappings between solutionand problem spaces.

o The Functional Ecology Model [Bergman et al, 2002]


18/36

Massimo Felici2004

CSC2004


o The Functional Ecology Model describes solution-

problem iterations

o The formal (i.e., formal logic) extension of thefunctional ecology model provides a framework tomodel and capture requirements evolution

SolutionProblemSolution

LocalSolution Space

Problem SpaceFuture

Solution Space

Objective

Requirements

Constraining

RequirementsRequirementsSpecification


19/36

Massimo Felici2004

CSC2004


o Example A Reparable Clock Design

Clock

BrokenClock

Tick

Break

The clockshould bereparable

None statetransactionis terminal

Clock

BrokenClock

Tick

Break

Repair


20/36

Massimo Felici2004

CSC2004


o Example A Tick-tock Clock Design

Clock

BrokenClock

Tick

Break

Repair

The clockshouldTick-tock

Clock 1

BrokenClock

Tick

Break

Repair 1

Clock 2

Tock

Repair 2


21/36

Massimo Felici2004

CSC2004


o The formal framework captures RequirementsEvolution in terms of solution space transformationso Requirements evolution is a co-evolutionary process.

Requirements evolution consists of the RequirementsSpecification Evolution and the Changes Evolution.

o Requirements evolution represents organizationalknowledge. It is part of organizational learningprocess.

o Solution and problem spaces are socially shaped.o Issue: the social-shaping of solution and problem

spaces is error-prone

o Any solution requires the organization commitmentin terms of resources


22/36

Massimo Felici2004

CSC2004


o Other requirements evolution exampleso Cascade effect - Changes propagate through the

mappings between solution and problem spaces

o Allocation of Safety Requirements in subsequentreleases

o Hybrid Solution Spaces System Models andHuman Cognition models


23/36

Massimo Felici2004

CSC2004


o Evolutionary dependency identifies how changeseventually propagate through emergent requirementsdependencies

o The formal extension of the functional ecology modelcaptures requirements evolutionary dependencies (interms of solution space transformation)

o Solutions are requirements dependencies

o The formal framework used at an higher levelof abstraction than design


24/36

Massimo Felici2004

CSC2004


o Capturing Evolutionary Dependency

F1Software

Architecture

F2Interface

Subsystems

F8System

Maintenance

Tooexpensiveandrisky!!!

/ /$

/ /$

/ /$


25/36

Massimo Felici

2004

CSC2004

A Taxonomy of EvolutionEvolution as Dependability


26/36

Massimo Felici2004

CSC2004

A Taxonomy of Evolution [1/2]

o It is possible to identify an

evolutionary space forsocio-technical systems

o Temporal Dimension

o Evolution in Design

o Evolution in Use

o Physical Dimension

o Hard Evolution

o Soft Evolution

Evolutionin Design

SoftEvolution

HardEvolution

Evolutionin Use

Temporal dimension

Physical

dimension


27/36

Massimo Felici2004

CSC2004

A Taxonomy of Evolution [2/2]o The identification of a broad spectrum of evolutions

in socio-technical systems points out strong

contingencies between evolution and dependability

o The better our understanding of socio-technicalevolution, the better system dependability

Software Evolution

Architecture (Design) Evolution

Requirements Evolution

Socio-technical System Evolution

Organisation Evolution

Evolutionin Design

SoftEvolution

HardEvolution

Evolutionin Use


28/36

Massimo Felici

2004

CSC2004



29/36

Massimo Felici2004

CSC2004


o On Requirements Evolutiono Socio-technical grounds of Requirements Evolution

o Empirical analyseso Capturing Requirements Evolution

o Observational Models of Requirements Evolutiono A bit of theory

o Evolution as Dependability


30/36

Massimo Felici2004

CSC2004

Requirements Evolution [1/2]

1. Heterogeneous engineering stresses a holistic viewpoint that

allows us to understand the underlying mechanisms of evolutionof socio-technical systems.2. Requirements, as mappings between socio-technical solutions

and problems, represent an account of the history of socio-technical issues arising and being solved within industrialsettings.

3. The formal extension of the functional ecology model providesa framework to model and capture requirements evolution.

4. The application of the proposed framework provides furtherevidence that it is possible to capture and model evolutionaryinformation about requirements (e.g., evolutionarydependency).

5. The discussion of scenarios of use stresses practical

necessities for methodologies addressing requirementsevolution.


31/36

Massimo Felici2004

CSC2004

Requirements Evolution [2/2]

6. The identification of a broad spectrum of evolutions in socio-

technical systems points out strong contingencies betweensystem evolution and dependability.

7. This presentation argues that the better our understanding ofsocio-technical evolution, the better system dependability.

8. This presentation is concerned with software requirementsevolution in industrial settings. It introduces methodologies to

empirically investigate and model requirements evolution, henceObservational Models of Requirements Evolution.

9. The results provide new insights in requirements engineeringand identify a foundation for Requirements EvolutionEngineering.

10.This presentation addresses the problem of empiricallyunderstanding and modelling requirements evolution.


32/36

Massimo Felici

2004

CSC2004

Observational Models of

Requirements Evolution-

Theory and Practice

Massimo FeliciLFCS, School of Informatics, The University of Edinburgh

[email protected]

http://homepages.inf.ed.ac.uk/mfelici/

The Critical System Conference29th-30th September 2004


33/36

Massimo Felici

2004

CSC2004

Background Slides


34/36

Massimo Felici2004

CSC2004

o The socio-technical grounds of requirements (evolution)o Systems Approach: system builders include heterogeneous

components, such as, mechanical, electrical, and organizationalparts, in a single system.

[Hughes and Hughes, 2000]

o Heterogeneous Engineering: People had to be engineered, too persuaded to suspend their doubts, induced to provide resources,

trained and motivated to play their parts in a production processunprecedented in its demand. Successfully inventing thetechnology, turned out to be heterogeneous engineering, theengineering of social as well as the physical world.

[MacKenzie, 1990]

o Social Shaping of Technology: The mechanisms underlying thesocial design and implementation of technology systems arereferred to as the Social Shaping of Technology (SST).

[MacKenzie, 1999]

On Requirements Evolution [1/3]


35/36

Massimo Felici2004

CSC2004

On Requirements Evolution [2/3]

o Requirements Evolution vs. Reuse and Maintenanceo Current requirements engineering methodologies provide

limited support to understand requirements evolution. Reuseand maintenance tend to focus on classifying and managingrequirements changes, rather than analyzing or evenanticipating changes.

[Lutz and Mikulski, 2003]o Requirements evolution involves the practice on analyzing

requirements changes in order to identify strategies dealingwith environmental changeso It is possible to classify requirements as stable or changing

[PROTEUS Project]o It is possible to identify the stable or volatile requirements

o Empirical analyses of requirements evolution are still patchyo Lack of support for analyzing evolutionary datao Requirements evolution (analysis) as work practiceo Organizational culture


36/36

Massimo Felici2004

CSC2004

On Requirements Evolution [2/3]o Modeling methodologies and languages (e.g., SCR, Intent

Specifications, etc.) advocate different design strategies.

Although these strategies support different aspects of(software) system development, they originate in a commonSystem Approach.

o Systems Approach: Practitioners and proponents embrace aholistic vision. They focus on the interconnections amongsubsystems and components, taking special note of theinterfaces among various parts.

[Hughes and Hughes, 2000]o Requirements change: About four year later the original TCAS

specification was written, experts discovered that it did notcover requirements involving the case where the pilot of anintruder aircraft does not follow his or her TCAS advisoryandthus TCAS must change the advisory to its own pilot. This

change in basic requirements caused extensive changes in theTCAS design, some of which introduced additional subtleproblems and errors that took years to discover and rectify.

[Leveson, 2000]

Date post:	07-Apr-2018
Category:	Documents
Upload:	wajidislam
View:	221 times
Download:	0 times

Observational Models of Requirements Evolution CSC2004 Light

Documents