Systems Theory

Systems TheoryIntroduction and Overview

Stefan Ratschan

Department of Digital DesignFaculty of Information Technology

Czech Technical University in Prague

Evropsky socialn fond Praha & EU: Investujeme do vas budoucnosti

Stefan Ratschan (FIT CVUT) MIE-TES 1 / 28

My Technical Data

Stefan Ratschan

Building A, room 1029

http://www.cs.cas.cz/~ratschan (Consultation Hours!)

[email protected]

Rest (and most) of the time:

Institute of Computer Science of theAcademy of Sciences of the Czech Republic

Pod Vodarenskou vez 2

Metro stanice Ladv


Introduction and Survey

Which study program at which university?

Mother tongue?

Who speaks/understands Czech?

Who is afraid of the word theory?

Who knows what UML means?

Petri net

Turing machine

Logic course?


What is a System?If you want to have some high-brow definitionwith references to Greek, Latin, Chinese philosophers:look up your favorite encyclopedia.

Computer systems: software systems, hardware systems, network systems

But: In the past, job of someone with a degree in computer science:

not any more!

and if you want an interesting andwell-paid job, it is not enough toknow about computer systems.


Todays and tomorrows computer systems (example)


Todays and tomorrows computer systems (example)

Interaction with

organizational systems

biological systems (cells, organisms, ...)

ecological systems

economical systems

social systems

Especially: trains, airplanes, cars, mobile phones

That is: complex technical devices

So: system =something complex that we encounter as computer scientists

Today there are computational devices almost everywhere,and a computer scientist should be able to handle this.


Your Systems?

Usage for examples throughout the semester

Bachelor thesis?

Other systems?


PolymathHowever, nobody has the slightest chance to be an universal expert


No need for being a polymathInstead: theory

Liquids always flow downwards

Abstract description and explanation of reality

Abstract: independent of concrete situation,can be applied in many situations (an advantage!)

Scientific theories

Objective: independent of a concrete person/place/time,can be applied by anybody/anwhere/anytime

Systems Theory: description and explanation of systems behavior,that is valid

for everybody, everywhere, and all the time.

Differently from HTML constructs, Java library etc. This is a university!Stefan Ratschan (FIT CVUT) MIE-TES 10 / 28

Goal of Course

Provide orientation in a world of systems

How possibilities does mankind have for describing systems?

organizational systems

biological systems (cells, organisms, ...)

ecological systems

economical systems

social systems

technical and computer systems

Scientific theories: Principles that hold

for everybody,

everywhere, and

all the time.


Characteristics of Complex Technical Systems

Human-built

Highly complex: design teams

Huge: > 1000 for aeroplanes

Diverse: Mechanical engineers, software engineers, control engineers,chip designers, etc.

Often safety critical

Reactive: continuous interaction with environment

Concurrent: many functions in parallel

Sometimes: dynamic structural changes (e.g., new cars added)

Interaction of physical and electronical components(cyber-physical system)


Physical vs. Electronic Components

More and more functions taken over by digital electronics(e.g., brake-by-wire, x-by-wire, x-by-wireless)

Design costs of electronics more and more dominating costs ofphysical system


Resulting Control Loop

Analog-to-digital (A/D) converter


How to Design Such Systems?

Example: train

Highly complex: first decompose into components

Which ones?

traction

train body

signalling

electricity

air conditioning

lighting

doors

on-board information, communication, and entertainment

tilting system


Further Scenario (simplified V model)

we want a train

we want tractions signalization etc.

design traction, signalization etc.

test traction, signalization etc.

compose the parts

test the whole train

What will happen?

If engineers do not start to fight(they cannot talk to each other),

if they produce anything reasonable at all(sub-systems still immensely complex),

if they resulting designs can be put together (each sub-system takesdifferent assumption about other sub-systems),

if the resulting train runs at all, it will crash on the track.Stefan Ratschan (FIT CVUT) MIE-TES 16 / 28

ProblemsWe see and check the whole product only at the end!

Movie: Tacoma Narrows bridgecollapse

V-model only fancy trial-and error

We want

detect problems already at the beginning

if possible, automatically

How to achieve this?Stefan Ratschan (FIT CVUT) MIE-TES 17 / 28

ModelAbstract system representation

Abstract: Partial description that hides details (i.e., hides complexity) and describes only those aspects of a system that

are important for the task at hand.

For example: We are looking for the mensa. Is this useful??


Hierarchy of System Models Informal requirements:

Train should follow signalling standard for Czech tracks (Partially) formal specification (but still: What do we want?) High level operational models (How should it work? e.g., finite state

automaton, Scilab demo) VHDL, C program Circuit design, machine code . . .

Hierarchy of abstractions:higher levels hide complexity,lower levels provide more details

Cycle (simplification, idealization): model based design Create system model Check if o.k. (if possible automatically) If necessary use models with lower level of abstractionStefan Ratschan (FIT CVUT) MIE-TES 20 / 28

Result

Advantages:

we work with whole system during the whole design process

from the beginning we can analyze/debug whole system

tool support: automatic translation between models (compiler, code generator) simulation (demo?) testing formal verification

Therefore: problems, and errors can be found early

Therefore:


Model Classification

Formal/informal (using natural language, figures etc.)

Dynamical vs. static models

Adaptive vs. fixed structure models


Goal of Course (practical version):Basic situation:

We want to design a system (software, train, mobile application etc.)that fulfills its specification

The method of trial and error is too expensive and not reliableenough.

Hence, before building the system, wedesign a model and analyse it automaticlly

Fancy big question:

How can we describe and computationally analyze complex systems?

Concrete goal: Learn formalisms, techniques, and tools formodeling and analysis of complex technical systems, such that

you can function in such a design process, and

you will produce high-quality systems.


Preliminary List of Lectures

1. Introduction and Overview

2. Basics of Practical Logic

3. Communicating Finite State Machines I

4. Communicating Finite State Machines II

5. Specification: Temporal Logic

6. Bounded Model Checking

7. Boolean Satisfiability Checking

8. Unbounded Model Checking

9. Petri Nets

10. Timed Automata

11. Continuous, hybrid systems

12. Hybrid sytems: verifications

13. Probabilistic models


SourcesSlides on Edux

I will not follow a certain textbook.

On the slides of the individual lectures I will provide references to sources

Lecture notes on Edux (under development)

You will not need them if you attend the lectures

If you miss some lectures, you will need additional sources

Lectures slides will not contain all information,I will draw examples, figures etc. onto the whiteboard

Be careful with wikipedia!

A large part of the lecture material is not there. If it is there, it often contains mistakes, bad explanations etc.

Wikipedia cannot replace understandingStefan Ratschan (FIT CVUT) MIE-TES 25 / 28

AssignmentsThe assignments will not be about

blindly applying some already known method to new data

Real life is never about blind application of known methods.

For solving assignments you will have to

understand the description of the problem

read, understand, and use the corresponding definitions

solving a new problem.

This you will need throughout your life.

Everything you need, I will present in the lecture(including some examples).

In assignment class you should be able to answer questions like:

According to which definition does this hold?

How did you use this definition?


Final ExamYou will be allowed to use any book, printouts of slides etc.

o.k., so Ill just relax during the semester,at the exam I will anyway be able to find the answers in books

This will not work!

The final exam will consist of two types of examples:

Examples of the same or similar typeas the assignments during the semester.

Examples, that test whether the student is able to apply the theoryfrom the lecture on his/her own in a new context

Time:

will be easily enough for well prepared people. People, who will start to list in slide printouts at the exam,

will have problems with time.Stefan Ratschan (FIT CVUT) MIE-TES 27 / 28

Next Lecture

Basis of all models: Logic

Logic is very often taught as a subjectdescribing the foundations of mathematics and computer science(Hilbert calculus, resolution algorithm and so on).

Which is great, but as a result,many students lack the ability to apply logic in everyday life

Next lecture: Basics of practical logic

From the theoretical point of view nothing new.

If are fluent in logic and its usage, no need to come(lecture notes, Chapter 1 and 2 without 1.9 and 2.2)


Documents

Systems Theory