Systems Tutorial - The Fundamentals
0 likes1,514 views
This document details a public lecture by Professor Peter Cochrane on the principles of systems, their complexities, and the interactions between natural and designed systems. It discusses the implications of system design in a world increasingly characterized by complexity, emphasizing the importance of understanding these systems to avoid mistakes and ensure positive outcomes. The lecture challenges conventional thinking and advocates for a comprehensive approach to studying and designing systems within interconnected environments.
![W hat ’s in THE BOX ?
What are the limits to what we describe and define
s(t) h(t) o(t)
What can we describe and define
Optical
Acoustic
+++
+++
Life forms
o(t) = h[s(t)] = h(s) for ease of notation
o = a + bt + ct2 + dt3 et4 + ft5
is the largest polynomial we
can solve for very limited and
narrow range of cases
In the absence of a closed form solution we often reduced to using polynomial or some
other form of approximate descriptor](https://image.slidesharecdn.com/systems1-180420152016/85/Systems-Tutorial-The-Fundamentals-39-320.jpg)
Systems Tutorial - The Fundamentals
- 1. Wed 28 March 2018 16:00 – 18:00 Out of hours public lecture Presented by Professor Peter Cochrane OBE Ipswich Waterfront Building Animated tutorial style with, demonstrations, videos & provoking propositions Organised and hosted by the UoS Innovation Centre Systems 1.0 Everything you needed to know about the subject before you started the course ! A detailed look at some of the fundamental principals and key differences between natural and designed systems on which we all dependant
- 2. "Simple can be harder than complex. You have to work hard to get your thinking clean to make it simple" Steve Jobs "Global terrorism is extreme both in its lack of realistic goals and in its cynical exploitation of the vulnerability of complex systems" Jurgen Habermas "Today the network of relationships linking the human race to itself and to the rest of the biosphere is so complex that all aspects affect all others to an extraordinary degree. Someone should be studying the whole system, however crudely that has to be done, because no gluing together of partial studies of a complex nonlinear system can give a good idea of the behavior of the whole" Murray Gell-Mann
- 3. Definitions & AXIOMS Can we describe what we mean by a system ? This singularly simple question directs us to an ongoing philosophical debate that is less than helpful for us as engineers and scientists We need something both meaningful and concise that affords us sufficient licence to assist our understanding in designing and building machines and networks that invoke positive advantage and progress at minimal risk employing progressively less energy & material
- 4. “A group of interacting, interrelated, or interdependent elements forming a complex whole” The most concise definition I can find but not entirely satisfactory for our purposes… literature seaRch Thousands of references/discourses on the topic * The whole may be simple, complicated or complex, and there may or may not be any interdependence !
- 5. A functionally related group of elements, especially: - The human body regarded as a functional physiological unit - An organism as a whole, especially with regard to its vital processes or functions - A group of physiologically or anatomically complementary organs or parts - A group of interacting mechanical or electrical components - A network of structures and channels, as for communication, travel, or distribution - A network of related computer software, hardware, and data transmission devices Correct but inconcise, incomplete, and unsuitable: literature seaRch Thousands of references/discourses on the topic
- 6. An organised set of interrelated ideas or principles -A social, economic, or political organisational form - A naturally occurring group of objects or phenomena: the solar system. - A set of objects or phenomena grouped together for classification or analysis - A condition of harmonious, orderly interaction - An organized and coordinated method; a procedure Correct but inconcise, incomplete, and unsuitable: literature seaRch Thousands of references/discourses on the topic
- 7. ‘A system takes energy, matter, information, and transforms their nature’ * Ergo: All Systems are Entropic Not Published My tight and sufficient (?) definition
- 8. A W IDER LENS Only art, science, and engineering Divisions: Mathematics, Physics Chemistry, Biology, Art are all artificial silos inflicted sometime after the reformation and whilst accelerating our progress & understanding they now sees widespread limited thinking and a lack of mutual understanding that is disadvantageous Galileo Galilei Michael Angelo Leonardo da Vinci
- 9. Taking an interest in every system known to mankind pays dividends in providing us with insights and challenging concepts and occasionally , really useful results... …we no longer design, deploy and operate our systems in isolation...we live in a world of natural and unnatural systems... evolved and designed... ...the way they connect coexist and interact is important especially when life dependency and mission critical issues are at stake ! JUST GOOD PRACTICE For completeness of enquiry we need a better radar
- 10. Systems are never stronger than their weakest element Systems are never simpler than their most complex elements Systems are always more complex than their most complex elements G E N E R A L A X I O M S For individual & connected/networked systems
- 11. • Complex systems never get easier to characterise • Simple systems tend to get more difficult to characterise • Complex Systems are never rendered simpler - without incurring costs ! • Simple systems tend toward being rendered more complex ! • Simple systems don’t make the complex simpler • Complex systems always make the simple more complex • There are no simple solutions to complex problems • There is a huge difference between complicated and complex systems G E N E R A L A X I O M S For individual & connected/networked systems
- 12. • Simple systems tend to migrate toward complication • Complicated systems tend to migrate toward complexity • Complex systems can comprise simple and/or complicated elements • Entropy always follows the direction (1, 2, 3) and never the reverse • The converse of the construct (1, 2, 3) seldom occurs - if ever ! • Cluster of simple/complicated systems can become complex a whole Our lack of understanding never deterred us from exploiting anything, but we have often witnessed some pretty big mistakes in the process! G E N E R A L A X I O M S For individual & connected/networked systems
- 13. Mother Natures propensity for Simple Systems resulting in Complex Outcomes WHAT WE KNOW FOR SURE A natural & designed world complexity - simplicity inversion
- 14. Humans propensity for Complex Systems resulting in Simple Outcomes WHAT WE KNOW FOR SURE A natural & designed world complexity - simplicity inversion
- 15. DominantSystemTrend WHAT WE KNOW FOR SURE We cannot manage 21C societies with 17C thinking/systems
- 16. V I S I B L E T R E N D Irreversible progression to complexity DESIGNED EVOLVED Well behaved Out of control Well Understood Knowledge Gap TOOLS MODELS Physical Laws Emulation Low Combinatorics Certainty the Norm High Predictability High Combinatorics Uncertainty the Norm Emergent Behaviour Mathematics Simulations Plasma Biology Physics Weather Universe Genetics Ecologies Chemistry Proteomics Combustion Earthquakes Global Warming Immune Systems Quantum Mechanics National Security Search Engines Globalisation Management Mobile Nets Leadership Intelligence Economics Nano-Tech Bio-Tech Conflict M&A COMPLEX Telephone Car Engine Jet Turbine MRI Scanner Rocket Motor Air Conditioner Server Farm Computer Lap-top Tablet Radio TV COMPLICATED Long Bow Catapult Pulley Canal Mill SIMPLE Drill Lathe Bicycle Ratchet
- 17. syst e ms g e n e ra l it i es Sureties, actualities, challenges, opportunities Digital Analogue Hybrid Analogue//Digital Our knowledge base Our understanding Made by mankind Made by machine Our species survival Our planets survival Machine intelligence Symbiosis necessary Challenges to be addressed dominant remains a vital core ubiquitous & growing advancing and accelerating limited by our technologies vital to our survival and prosperity vital to our survival and prosperity depends upon good systems ***** depends upon good stewardship/tech overtaking us in many areas man-machine partnership underway ***** formidable but interesting and vital *****
- 18. General Observations All systems share similar mathematical/analysis frameworks Ergo: Electrical, Electronic, Mechanical, Civil Engineering & Physics share the same/very similar equations sets…’know one and you know the other’
- 19. Differences: Chemistry, Biology, Sociology, Information, Network Systems/ Engineering tends to observe through a different lens but share some similarities General Observations All systems share similar mathematical/analysis frameworks
- 20. ‘SIMPLEst ’ SYSTEMS Lie within the grasp of one human mind & hands Designed top down Always predictable Easy to specify Easy to design Easy to model Easy to realise Easy to control Easy to operate Do not evolve Do not adapt Unchanging Stable/well behaved Essentially Linear Laws of physics apply Mathematics works Linear behaviour Predictable Conceived, designed and produced by one human
- 21. S I M P L E S YS T E MS Within the span of ‘a’ human mind or team Designed top down Always predictable Easy to specify Easy to design Easy to model Easy to realise Easy to control Easy to operate Do not evolve Do not adapt Unchanging Stable/well behaved Essentially Linear Laws of physics apply Mathematics works Linear behaviour Predictable
- 22. Difficult to understand Designed top down Difficult to specify Difficult to design Hard to model Hard to realise Hard to control Hard to operate Seldom non-linear Generally predictable Do not evolve - stable Stable/well behaved Essentially Linear Laws of physics apply Maths ‘mostly’ works(ish) Unnatural materials required Computer modelling a necessity Complicated SYSTEMS A deep understanding beyond the grasp of ‘a’ human
- 23. Complicated
- 24. Initial/intuitive top down design Evolution rapidly dominates Modelling a major challenge Accurate specification hard Fundamentally stochastic Design; a tough challenge Realisation is easy/hard Operation is easy/hard Control is easy/hard Non-linear/chaotic Complex SYSTEMS Beyond human wisdoms and mental abilities Established wisdoms do not apply Mathematics non-contributor Emergent behaviours rule No general laws
- 25. C O M P L E x The concatenation of large numbers of simple and/or complicated sub- systems can lead to unpredicted non-linearities and thus surprises - unexpected emergent behaviours due to the whole becoming complex
- 26. M ac h i n e D ES I G N E D We are in (just) a realm where we have no clue! AI + AL derived AL evolution Impossible (?) to understand Seeded from the bottom up Accurate spec impossible Outcome unpredictable Beyond (?) modelling Very easy to realise No human control Autonomous code ‘Always’ non-linear Only modest compute power Good network connectivity Maths nears irrelevancy Now ‘breeding’ malware
- 27. Stability Two extremes/bounds Conditionally Stable Conditionally Unstable Only becomes unstable when there is some system failure Under any and all conditions
- 28. Output Input Linear : NON-LINEAR Confusing and often confused These are all non-linear responses as they give wildly different outputs for the same input every time These are all linear responses in that an input gives the same output every time Established wisdoms do not apply Mathematics non-contributor Emergent behaviours rule No general laws
- 29. THESE DAYS HAVE Long GONE We a r e n o w m o s t d e f i n i t e l y i n a m a c h i n e a g e
- 30. In a simple disconnected world... ...we can address problems in isolation ...and simple solutions mostly work In a complex connected world... ...we have to consider the whole ...simple solutions never work AI ...modelling, simulation, decision support ...BIG DATA ...everything online, and networked ...a living organism... R E A L I T Y Simple is now a rarity AL …evolved designoid solutions
- 31. BIGGEST WORRY The majority do not comprehend Simple linear and a well behaved arena: intuition, experience, & wisdoms mostly worked well and were adequate across the most societies A universe characterised by a limited range of relationships, concepts and equations that were readily understood Non-linear, complex and highly unpredictable arena: intuition, experience, wisdoms are mostly unreliable at best & highly dangerous at worse A universe characterised by a wide range of immature and developing concepts/computer models not readily understood by lay people, politicians, +++ PAST TO DAY FUTURE Economics & market forces are failing Ignorance and bigotry on the rise Politics & democracy is in peril
- 32. LEMMING LIKE Doing what they thing is right
- 33. Compounding
- 34. Fourier + Laplace - time and frequency g(t) f(t)❋g(t)f(t) F(ω) G(ω) F(ω).G(ω) Simple Multiplication Complicated Convolution Laplace forces integrals to converge - best used on transients signals Fourier can see integrals to diverge - best used on repetitive signals BASIC SYSTEM Nomenclature - Methodology Frequency Domain Descriptor Time Domain Descriptor This is the original analogue version we have now rendered discrete and digital - Digital Fourier/Laplace Transform
- 36. Temporal Function Frequency Function Frequency Function Temporal Function Applicable to time and s p a c e , h e a t fl o w , mechanical systems, probability theory +++ TRANSFORMATIONS Difficult problems rendered easy(ier) Fourier: time frequency domain = Laplace: time frequency domain transient signals
- 37. T RA NS FO R M AT I O NS Analogue and Digital/Discrete) formulations
- 38. ENVIRONMENT? s(t) h(t) o(t) Other systems of the same or differing type may be sharing the same space or some part of it, and therefore there can be many obvious and hidden opportunities for aliasing.... Air Water Earth Machines Lifeforms Fluids Solids Chemicals Radiation Information Chemical Physical Information/Data Processing Mathematical Natural Unnatural Biological Electrical Electronic Mechanical Computational Optical Acoustic Organic Inorganic Life forms +++
- 39. W hat ’s in THE BOX ? What are the limits to what we describe and define s(t) h(t) o(t) What can we describe and define Optical Acoustic +++ +++ Life forms o(t) = h[s(t)] = h(s) for ease of notation o = a + bt + ct2 + dt3 et4 + ft5 is the largest polynomial we can solve for very limited and narrow range of cases In the absence of a closed form solution we often reduced to using polynomial or some other form of approximate descriptor
- 40. AND the output ? s(t) h(t) o(t) In the general case it impacts/changes the environment and the input and is often a grossly non-linear series of loops e(t) f(t)
- 41. FEEDBACK & FEEDBACK ? s(t) h(t) o(t) FB o(t) FF s(t) Noise reduction loop -ve FeedBack induces stability +ve FeedBack induces instability - an oscillator Mathematical tractability reduced by the # loops Stability and response shaping loop
- 42. Fourier + Laplace - time and frequency g(t) f(t)❋g(t)f(t) F(ω) G(ω) F(ω).G(ω) Simple Multiplication Complicated Convolution Laplace forces integrals to converge - best used on transients signals Fourier can see integrals to diverge - best used on repetitive signals BASIC SYSTEM Nomenclature - Methodology Frequency Domain Descriptor Time Domain Descriptor This is the original analogue version we have now rendered discrete and Digital Fourier/Laplace Transform
- 43. General SYSTEM traits s(t) h(t) o(t) s1(t) s2(t) s3(t) si(t) o1(t) ok(t) o3(t) o2(t) hi(t) Simple Singular Linear Complicated /Complex Multi - I/O Linear Non-Linear In general can be fully tested and characterised In general cannot be fully tested and characterised
- 44. All known, understood, well described and characterised, bounded, and well behaved with causality preserved Contained/bounded in/by some known, or well defined, environment/conditions Simple System - Key Features 1 s(t) h(t) o(t) s(t) = Stimulus h(t) = Operator o(t) = Output } s(t) and o(t) originate and terminate within the environment
- 45. Response matches need Symbiotic with the environment Predictable, reliable, with a fast recovery time Upgrades and changes not traumatic or risky Shocks are not terminal or unduly debilitating Reproducible, easy to deploy and maintain/repair/replace Simple System - Key Features II s(t) h(t) o(t) Sometimes we cannot satisfy this wish list 100%
- 46. All known, understood, well described and characterised, bounded, and well behaved with causality preserved Contained/bounded in/by some known, or well defined, environment/conditions Complicated/Complex - Features I s(t) = Stimulus h(t) = Operator o(t) = Output } s(t) and o(t) originate and terminate within the environment s1(t) s2(t) s3(t) si(t) o1(t) ok(t) o3(t) o2(t) hi(t) X Any one or more or all of these conditions may no longer true X X
- 47. s1(t) s2(t) s3(t) si(t) o1(t) ok(t) o3(t) o2(t) hi(t) One or more of these conditions no longer holds true Response matches need Symbiotic with the environment Predictable, reliable, with a fast recovery time Upgrades and changes not traumatic or risky Shocks are not terminal or unduly debilitating Reproducible, easy to deploy and maintain/repair/replace Almost by definition we cannot satisfy this wish list 100% X Complicated/Complex - Features II
- 48. Man and Mother Nature… Design, understanding, desire, intent v evolution and chance Only we design Only we optimise Only we do centralised control Only we comprehend and assume responsibility Design v evolution Top down v bottom up Optimisation v good enough
- 49. Mother Nature…proviso Only evolves systems Only builds bottom up and never top down Only goes for ‘good enough’ and optimises nothing She conceals her underlying complexity at every level of her constructs and activity... Extremely complex cells and reasonably simple constructs with distributed control
- 50. Hundreds of diverse inputs and outputs: cannot be fully flood, or combinatorially tested… Hundreds/thousands of feedback & feedforward loops along with memory and adaptation… R EA L ITY B UT ES ! Complexity in all things becoming the norm
- 51. “Perfection is the enemy of Good Enough” Defining ‘good enough’ is not always trivial and is generally the biggest challenge ! ~80% of the need satisfied by ~20% of the effort….and then often destroyed/devalued by specification creep…. APEING Nature We are building evolved systems AI, AL, IoT, Internet, Security are now evolutionary This future will be full of surprises - ‘emergent behaviours’ - and we are having to abandon the idea of 100% testability and characterisation - and we will surely lose control too! New mind sets and new system thinking is going to be essential
- 52. U N R E A L Not realistic/impossible
- 53. T R A D E O F F S You can’t have it all - not forever anyway
- 54. Size Scale Complexity Connectivity Sophistication Connectivity MTBF Speed Agility Reliability Testability Predicability Responsivity Cost MTTR Latency Power Heat Resources Often difficult to define with any great precision Common/General system traits
- 55. A long storage life and very short operational activity HUH ??
- 56. Short storage life and very long operational activity HUH ??
- 57. Brittleness ALWAYS rules Reliability/resilience and optimisation are mutually exclusive Highly optimised components result in mission critical failures
- 58. η Failure free operating cost with efficiency f1(𝓷) Failure cost with efficiency f2(𝓷) € € € Efficiency 𝓷 f1’(𝓷) + f2’(𝓷) = 0 For the specific exponential case: C = Aexp(-a. 𝓷) + Bexp(b. 𝓷) 𝓷o = b - a __1 loge Aa Bb _ The optimum operating efficiency is found to be: Efficiency is ‘easy’ People overlook the cost of failure!
- 59. Bathtub Reliability Curve Failure Rate End of life wear out failures Normal/useful life low level failures Time Infant mortality - early failures Failure Rate
- 60. 0.001 0.01 0.1 1 10 100 1,000 10,000 100,000 1,000,000 1 10 100 1,000 10,000 100,000 1,000,000 MTBF - hours Mean Time Between Failures MTTR - seconds Mean Time To Repair 99.999% 99.99% 99.9% 99% 90% 0.1 1 year 10 years1 month 1 week 1 day 1 minute 1 hour 1 day Availability Offsetting inevitable failures MTBF = 11.4 years 1 second
- 61. System 1 System 2 System N ? Not co-located Not the same power feed Not the same software Not the same network connections No centralised control Switch Or Sum Voting A =1- (1 - Av)N Av Av Av A = Availability Hot standby Offsetting inevitable failures
- 62. MTBF = Mean Time Between Failures MTTR = Mean Time To Repair Mean = Average - and Averages don’t tell us much! Distribution is key ! Availability = Av = MTBF/(MTBF + MTTR) Unavailability = Uv = 1 - Av = 1 - MTBF/(MTBF + MTTR) = MTTR/(MTBF + MTTR) Relationships At the most basic ‘average’ level
- 63. Resilience = The capacity to recover Failure modes: Graceful/Phased Predominant in the world of analogue system - precursors evident
- 64. Resilience = The capacity to recover Failure modes: catastrophic/instant Predominant in the world of digital systems - precursors often hidden
- 65. SOme still to do’s Specifically for the digital domain Failure precursor detection Integrates graceful fail mode Pre-emptive failure switchover Golden ‘brick’ reference models Comprehensive test methodologies Brittleness threshold/onset detection Abnormal system operation identification No-GO realm/region/mode recognition Autonomous failure recover modes Penetration/illegal act detection Full outcomes characterisations Decision ‘post mortem’ facility Species comparators ++++
- 67. Sniper Locating M u l t i p l e a c o u s t i c s e n s o r s
- 68. Sniper Locating with acoustic sensorsAcoustic Cone of U n c e r t a i n t y 3 - 5o Sensor Unit War Zone Ambient Noise Dynamic R a n g e Limited Today’s Performance ~200m
- 69. Acoustic Cone of U n c e r t a i n t y 3 - 5o Sensor Unit Dynamic R a n g e Limited ~200m In Combat Multi-Mic ArrayMODERN SYSTEM
- 70. Filtering Post Microphone analogue/digital filtering/ processing ~~~ DSP ~~~ Noise Filtering Directivity - Focus Voice Characteristics Anticipated Subject Matter Correlated sniper shot signal - a matched filter by any other name Dynamic range limiting background noise T i m e waveform and spectrum o f s i n g l e sniper shot
- 71. Primary dynamic range limiter - the diaphragm Diaphragm driven to mechanical saturation by noise and thereby limiting dynamic range for the wanted/weak signal T i m e waveform and spectrum o f s i n g l e sniper shot Pre-Filter
- 72. Noise Filtering Directivity - Focus Voice Characteristics Multiple analogue resonant cavity filtering Pre-Diaphragm acoustic filter reduces noise input to improves dynamic range and directivity T i m e waveform and spectrum o f s i n g l e sniper shot Filter Config
- 73. Filter Characteristic A wide range of selectivity configurations is possible Naked Microphone Insert Filter 1 Filter 2
- 75. E a c h f i lt e r/m i c a s s e m b l y i s m e c h a n i c a l l y isolated by foam - n o a c o u s t i c linkage assembly to assembly Detail
- 77. Sniper rifle
- 78. Ambient Battlefield Noise Sniper Rifle Spectrum Containment > 95% of Sniper Spectral Energy Filter Rejection ZoneThis region is not generally critical to detection but can contain energy vital to identification PeaktoPeakhighly distancedependent 2 10 100 1k 5kHz
- 79. Ambient Battlefield Noise Sniper Rifle Spectrum Containment > 95% of Sniper Spectral Energy Filter Rejection Zone PeaktoPeakhighly distancedependent 2 10 100 1k 5kHz Relative and non specific energy levels for the purpose of illustration - all location and situation defined Prototype Filter Response
- 80. Sniper rifle buried in machine gun signal
- 81. PRE-MICROPHONE ACOUSTIC FILTERS WITH GAIN Sniper rifle detected with pre-microphone filter detection
- 82. Sniper Rifle + Machine Gun Autocorrelation Sniper Detected Pre mic acoustic filtered signal Processing Delay
- 83. “Anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that 'my ignorance is just as good as your knowledge” ― Isaac Asimov ENGINEERING CHALLENGE We face a tidal wave on skepticism based on ignorance “This a ‘systemic failure’ and it is beholden to all the professions and educationalists to counter this trend through the clear and justified provision of the raw facts and truth of a situation or prospect - truth and knowledge are the foundation of our technological society - they are vital to our survival as a species”
- 84. Any further questions or thoughts ?? cochrane.org.uk