COMPLEX SYSTEMSENGINEERING: THE FUTURE OF

SYSTEM DESIGN

Prof. Rashmi JainDepartment of Industrial and Systems Engineering 

National University of SingaporeNovember 18, 2011

Email: iserashm@nus.edu.sg

Topics for discussion

• Complexity Challenges• Reductionist VS Systemic View• Complexity of Embedded Software Systems (EWS)

• Systems Engineering Challenges• Design For Complexity – The “Right Design”

So What is Engineering?

"design under constraints" ‐ where the constraints include the laws of nature, cost, safety, reliability, environmental impact, manufacturability, and other factors.

Engineering in K‐12 Education: Understanding the Status and Improving the Prospects, 2009 by the National Academy of Engineering and the National Research Council's Center for Education.

Systems Engineering

“An engineering approach that transforms anoperational need or market/societalopportunity into a system solution descriptionto (that would) support detail design, itsdevelopment, production, maintenance,retirement and obsolescence by focusing on itsstakeholders – the end users, operators,maintainers, financiers etc.”

Complexity – The Grand Challenge• Any natural, or human‐designed complex system is, at least 

partially, unpredictable. • Examples of such complex unpredictable systems include human 

beings, complex telecommunications networks, embedded software systems, computer operating systems, large organizations, road traffic, complex manufacturing processes, ecosystems, markets, weather etc.

• Automation is not the solution: Computers are bad at managing unpredictability:– robots are unable to autonomously navigate through unpredictable 

natural environments; – computerized manufacturing systems fail to emulate the dexterity 

and flexibility of human craftsmen; – computer and telecommunications systems have only a limited 

ability to defend themselves against external attack or internal malfunction, 

– computerized customer support systems are unable to interact effectively with users. 

Complexity Theory

• Promises exciting possibilities for understanding the natural, social and other man‐made systems.

• Acknowledges the interrelated nature of things and of “emergence” phenomena –whole is experienced as greater than the sum of its parts. 

• Emergence – Spontaneous self organization –not predictable

Complexity: From Reductionism to Holism….• Reduction phenomena looks at entities as detached real things, behaving according to fixed causal relationships.

• Reductionist thinking directs people to seek solutions in terms of causal factors – A caused B, rather than through systemic awareness.

• These have had a direct impact on how social rules, laws, organizational and social practices have evolved and  the emerging social behavior.

• A Systemic Awareness demands a more complex effort at understanding the situation ‐ The role of interactions of entities, patterns of such interrelationships and emerging difficulties due to many unknowns of the situation.

System Behavior• Systemic Approach argues that system behavior is more meaningfully  understood as the result of feedback loops where variables are interrelated

• Dynamic system behavior is capable of producing unexpected variety and novelty through spontaneous self‐organization. Spontaneous means that which emerges is not predictable

• Emergence is unpredictable because it results from details of dynamics that are inherently unknowable to the human mind – An extreme example of “Amazon Butterfly” 

System complexity measures…a view• Complexity is a measure of how hard something is to understand or achieve– Components —How many kinds of things are there to be aware of?– Connections —How many relationships are there to track?– Patterns — Can the design be understood in terms of well‐defined patterns?– Constraints — Timing, precision, algorithms

Complexity arises in situations where “an increasing number of independent variables begin interacting in 

interdependent and unpredictable ways.”

Systems of Increasing Complexity 

Systems are increasingly getting complex:• Flight simulators• Mobile communication systems• Embedded software systems• Aero‐engines• Commercial aircraft• High‐speed trains• Intelligent buildings• Baggage handling systems• Automated printing press• Control systems: air traffic, railway, 

electricity, telecoms, global finance• Projects: Millennium dome, Channel Tunnel 

Rail Link, Boston’s Big Dig Project.

Source: Product Complexity, Innovation and Industrial Organization; Mike Hobday (1998); Research Policy (26, 1998). 

# of “customers"# of functions

# of users# of disciplines

# of sub-systems

Life cycle: •Operations•Logistics

•Maintenance •Retirement

As product complexity increases, additional concepts, tools and methods and modes of thinking need to be deployed: Need for a new approach to

managing complexity

11

Embedded Software (ESW): Increasing Software Content in Automobiles

Source: MercedesMechanics + Electronics + Software

12

Integration for Cost, Quality and Packaging

# of Functions

1990 2000 2005 2010

# of ECUs *

Integration‐ Cost‐ Reliability‐ Packaging

* Electronic Control Units

13

Growth in Vehicle ECU Market

*US$36.8 billion (US 13.2)

*US$52.1 billion (US 16.9)

Source: Accenture Study, Tuning into Tomorrow’s Frequencies: How Product Development in Automotive Electronics Drives High Performance, 2005* Revised Estimates: http://www.instat.com/press.asp?Sku=IN0603375RE&ID=1752

14

Growth in Vehicle ECU Market• Growth forecasts across all product segments and regions ‐

strongest from the emerging markets in Asia and East and Central Europe. 

• North America is a mature market for a number of automotive electronic systems – still strong growth will be seen for electronic braking, steering and driver.

• Powertrain electronics accounted for 32%  (US$18.5 billion) and overall safety and convenience accounted for 50.3% of the global market in 2005. 

• Body/chassis electronics will show an average growth of 9.4% per year between 2005 and 2010 increasing to US$6.3 billion.

15

ECUs/ESW: Managing Consequences of Design and Resulting Integration Complexity

• Most Innovations in Automotive (Drive‐by‐wire, Driver Assistance...) based on Electronics and Software

• Increasing Complexity:  Growing Functionality and Interaction between Subsystems 

• Cost Pressure– Increasing Warranty 

Cost– Increasing Cost of 

Quality Assurance

Seventy-seven percent of respondents believe that over the next three years, ESW will become even more important

to their products.*

* Accenture Study: The Embedded Software Industry: Challenges and Successes, 2006

16

Trade‐Offs: Features VS Price 

Source: Accenture Study, Tuning into Tomorrow’s Frequencies: How Product Development in AutomotiveElectronics Drives High Performance, 2005

17

Managing Complexity: Integrate vs. Plug‐and‐Play

Source: Accenture Study, Tuning into Tomorrow’s Frequencies: How Product Development in AutomotiveElectronics Drives High Performance, 2005

18

Challenges for Achieving Embedded Software Company Objectives: A Survey

Source: Accenture Study: The Embedded Software Industry: Challenges and Successes, 2006

19

Changing Nature of ESW Development

Source: Accenture Study: The Embedded Software Industry: Challenges and Successes, 2006

20

ESW Impact on Product Development

Source: Accenture Study: The Embedded Software Industry: Challenges and Successes, 2006

21

Realities of the ESW• Hardware companies ‐ creating everything from mobile phones and automobiles to refrigerators to television sets ‐ encounter numerous challenges when integrating embedded software into their products.

• The growing complexity (on‐demand services) of such devices as mobile phones has created a number of challenges for ESW developers. 

• As devices powered by ESW have moved from isolation to connection, the inherent power and space constraints of their ESW have worsened.

• And as the devices perform greater functions, they require additional engineering expertise.

22

Increasing complexity as a result of increasing functionality and ECUs,

different types of interfaces, and different types of components.

Costs are driven by integration, testing, quality issues, and competitor

efficiency

Quality Issues: Due to poor testability. Results in increasing recalls and

warranty exposures

Integration and testing issues as a result of legacy design and

architecture (and inflexible in many ways), increasing use of ESW.

Involving multiple compliance requirements, standards, different levels

of technology maturity

Segment stakeholders and their needs.

What makes ESW different from other software development is resource

constraints. You have to think about space, power conservation and pixel

constraints. You have to balance power, simplicity, footprint, cost and

size.

Systems Engineering Challenges

Why Do Complex Systems Fail?• Complex systems fail not because they fail to accomplish their 

nominal purpose. Complex systems typically fail because of the unintended consequences of their design ‐ emergence.

• More system failures arise due to unpredicted consequences such as dysfunctional component interactions, human‐machine mode‐confusion and environmental factors. 

• A critical challenge for the future development of  systems is to build systems with a robust ability to manage unpredictability. 

• Currently, the only systems with the ability to achieve this goal are biological organisms and communities of biological organisms.

• System engineering fundamentally is concerned with understanding, simplifying, and minimizing, in a complex system, the unintended interactions between elements desired to be separate.

23

Biological OrganismsBiological organisms is based on design principles that differ profoundly from complex man‐made systems.1. Biological organisms exploit physics and chemistry. The functioning of biological organisms and communities leverage highly specific chemical and mechanical interactions which do not need to be explicitly programmed in their genetic code.2. Organisms grow and develop. A single piece of code (the organism’s genome) specifies the structure and function of millions of different cells. 

– During the development process the organisms acquires new information from the environment (e.g. memories or behaviors represented in the brain, the strengths of individual muscles). 

– The volume of this information is often much larger than the information explicitly represented in the genome. This makes both for flexibility (different environments produce different structures and behaviors) for compact code and for evolvibility (small changes in genotype can produce major modifications in phenotype).

3. New biological organisms are the product of evolution. The emergence of novel capabilities depends on relatively minor modifications to previous designs. In some circumstances (e.g. the evolution of homo sapiens from earlier primates) the changes may involve only small segments of the genetic code. 

24

Design For Complexity

25

Design for complexity

• From problems to issues and dilemmas• Laws of unintended consequences• Rationale for unintendedness• Sources of unintended consequences• The Right Design

26

The Complexity Triad: Moving from problems to dilemmas

Problem Solving

Issues

Systemic Awareness 

of Complexity

Dilemmas

Problem solvers  must grapple with complex interrelationships and emergent behavior. Focus on managing interrelated issues rather then the ‘process’ of problem solving

Debating interrelated issues aims to induce learning between people  through negotiations. Deep rooted dilemmas might come in the way of negotiations  

Issues rather than problems arise because of different views and experiences

Sources of Dilemmas based on individual 

experiences should be sensitized in negotiation

Pushing of consensus strenuously might be detrimental to the diversity of experiences and perspectives. Dilemmas should stimulate a thoughtful process of exploration

27

Understanding ‘Emergence’: Laws of Unintended Consequences• The Law of Unintended Consequences:

– actions always have effects that are unanticipated or "unintended." In the worst case, the outcomes are exactly the opposite of the purpose of the original actions. Often the resultant situation ends up being worse than the situation before the action was taken.

• This phenomenon has been around for centuries. In 1692 John Locke, the English philosopher spoke against a proposal to cut the maximum permissible rate of interest. Locke argued that instead of benefiting borrowers as intended the bill would punish the borrowers because people would find ways to circumvent the law and the costs of circumvention would be shifted to the borrowers.

• The most celebrated work on this topic originated from the Nobel LaureateFriedrich A. Hayek. His most significant contribution was to make clear “howour present complex social structure is not the result of the intended actionsof individuals but of the unintended consequences of individual interactionsover a long period of time, the product of social evolution, not of deliberateplanning."

• The same logic applies in understanding the nature of unintendedconsequences of system designs operating in complex environment and attimes expected to support very unpredictable and impossible scenarios.

28

Emergent Behavior• Understanding the impact of the unforeseen and unexpected interactions of the elements (some of which themselves be unexpected) in a system and its operational environment is the challenge of system designers in the inevitable and increasingly complex world that we have to design for.  

• The emergent behavior or nature of such systems creates the inability on our parts to understand the constraints and scope of system and its design. 

29

Basis of ‘unintendedness’ of consequences: Where do they come from?

• What is the basis of ‘unintendedness’ of consequences?  In other words – what are the sources of unintended consequences? What contributes to the “unintendedness” 

• Is it our inability (lack of ability) to foresee the resulting behavior of the system that bring these about OR is it the result of the interaction of the system elements that leads to totally new behavior – how can this be foreseen OR can it be?  

• These will give us insights into what kinds of things we should be considering based on what leads to or results in specific unintended consequences.  

“The law of unintended consequences is what happens when a simple system tries to regulate a complex system.”  ‐ Perceived Oversimplification

30

Basis of ‘unintendedness’ of consequences………• Should our focus be more on validation vsverification?  So rather then ensuring that the design meets the specs we focus on whether we identified the right specs in the first place or our approach was flawed.  

• Does our design address feasibility and testability? • How can we move the focus of design analysis more upfront rather than focusing on implementation and post‐implementation aspects?  

31

Sources of Unintended Consequences

• Robert K. Merton was an American sociologist who identified five sources of unanticipated consequences: – Ignorance It is impossible to anticipate everything, thereby leading to 

incomplete analysis– Error Incorrect analysis of the problem or following habits that worked in the 

past but may not apply to the current situation– Immediate interest, which may override long‐term interests– Basic valuesmay require or prohibit certain actions even if the long‐term 

result might be unfavorable (these long‐term consequences may eventually cause changes in basic values)

– Self‐defeating prophecy Fear of some consequence drives people to find solutions before the problem occurs, thus the non‐occurrence of the problem is unanticipated. Reverse Self fulfilling prophecy

• Much more work needs to be done in this area. Some thoughts have been proposed by Tenner on this topic of improving the understanding of unintended consequences of design.

Merton, R.K. Social Theory and Social Structure, Enlarged Edition, 1968 [1949], The Free Press, New York.32

The Right Design

33

The right design…..

• What is the ‘right’ design?  …….Design is a classic whole‐minded aptitude. It is a combination of utility and significance….stripped to its essence, it can be defined as the human nature to shape and make our environment in ways without precedent in nature, to serve our needs and give meaning to our lives [i].

• The idea of design and development is what most distinguishes engineering from science, which concerns itself principally with understanding the world as it is [ii].[i] John Heskett.; [ii] Petrosky, Henry, Invention by Design, 1996,

34

So how do we foresee the right design…..• Balancing the science and art of engineering in order to design successful systems –systems that take into account not just the obvious and intended functionalities but also provide for the unintended consequences of interacting system elements. 

• So how does one define this balanced approach of designing systems – getting the right design?

35

The right design…..• What is the ‘right’ design?  …….• The term ‘right’ has to be interpreted in the contextfor which the design is being made

• The scope defines the ‘right’ as well – clear boundaries– No engineering problem is ever solved to everyone’s satisfaction. Engineering is the art of compromise, (through trade‐offs) and there is always room for improvement in the real world.  

‐ Petroski• Innovation and creativity is an important element of the right design. How have you approached the problem differently (creatively)? 

36

Approaching the problem differently: Some guiding principles for a ‘right design’

• What’s unique about your approach? What is the valueproposition? What are its limitations?

• An approach guided by an honest and fair assessment of theconstraints – what can it not do? What are the pre‐conditions for it to be successful?

• How does the design ‘empathize’ with its stakeholders – theability to imagine yourself in someone else’s position and tointuit what that person is feeling. It is the ability to stand inothers’ shoes, to see with their eyes, and to feel with theirhearts.

• Understanding the relationships within relationships ‐improving our ability to design for decomposition, interfaces,integration, coupling, traceability, testability and other waysof implementing and controlling system functionalitythrough design.

37

• Understanding the dynamic nature of such relationships and the factors that influence this dynamism – factors such as technology maturity, knowledge (or lack of) about the context/environment, the foreseeable probable trajectory of emergence etc.

• Last but not the least, providing for the usability and understandability of the product and its features in the design. 

• The unintended consequence of failing to do this is human error. Humans do make mistakes, but with the right design, the incidence of error and its effects can be foreseen and minimized.

• Warning labels and large instruction manuals are signs of failures, attempts to patch up problems that should have been avoided by proper design in the first place.

Approaching the problem differently: Some guiding principles for a ‘right design’

38