Design and Architectures for Embedded Systems Prof. Dr. J. Henkel - - PowerPoint PPT Presentation

▶
design and architectures for embedded systems
SMART_READER_LITE
LIVE PREVIEW

Design and Architectures for Embedded Systems Prof. Dr. J. Henkel - - PowerPoint PPT Presentation

Aug 23, 2022 236 likes •812 views

ESII: Spec & Modeling PARTII 1 Design and Architectures for Embedded Systems Prof. Dr. J. Henkel Prof. Dr. J. Henkel CES CES - - Chair for Embedded Systems Chair for Embedded Systems University of Karlsruhe, Germany University of

slide-1
SLIDE 1

1 ESII: Spec & Modeling PARTII

Design and Architectures for Embedded Systems

  • Prof. Dr. J. Henkel
  • Prof. Dr. J. Henkel

CES CES -

  • Chair for Embedded Systems

Chair for Embedded Systems University of Karlsruhe, Germany University of Karlsruhe, Germany

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

University of Karlsruhe, Germany University of Karlsruhe, Germany

Today: Specification and Modeling PART II Today: Specification and Modeling PART II

  • Model and system properties

Model and system properties -

slide-2
SLIDE 2

2 ESII: Spec & Modeling PARTII

Where are we ?

System specification Design space exploration

  • low power
  • Performance
  • Area
  • …

System

  • models of computation
  • Spec languages

Estimation&Simulation

  • low power

Optimization

  • low power
  • performance
  • Area, …

refine

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

  • Emb. Software

Optimization for:

  • low power
  • Performance
  • Area, …

Embedded Processor Design

  • extens. Instruction
  • Parameterization
  • …

Integration Hardware Design

  • synthesis

Middleware, RTOS

  • Scheduling

System partitioning

  • low power
  • performance
  • Area, …

Tape out Prototyping

embedded IP:

  • PEs
  • Memories
  • Communication
  • Peripherals
  • …

IC technology

slide-3
SLIDE 3

3 ESII: Spec & Modeling PARTII

Outline

Motivation System and Model Properties Rugby Meta Model Case Study: A Design Project

(src: A. Jantsch)

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Case Study: A Design Project

(src: A. Jantsch)

slide-4
SLIDE 4

4 ESII: Spec & Modeling PARTII

What are models for?

=> => perform various tasks of a design process: perform various tasks of a design process: Performance modeling Functional modeling and specification Design and synthesis Validation and verification

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Validation and verification Test vector generation Test coverage analysis Architecture evaluation and mapping Technology mapping Placement and routing

(src: A. Jantsch)

slide-5
SLIDE 5

5 ESII: Spec & Modeling PARTII

What is a model?

Model: A model is a simplification of another entity, which can be a physical thing or another model. The model contains exactly those characteristics and properties of the modeled entity which are relevant for a given task. A model is minimal with respect to a task, if it does not contain any

  • ther characteristics than those relevant for the task.
  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

  • ther characteristics than those relevant for the task.

A model relates to an entity A model is a simplification of that entity A model is related to a task and an objective A model may relate to a not yet existing entity

(src: A. Jantsch)

slide-6
SLIDE 6

6 ESII: Spec & Modeling PARTII

Property of models

Inherent property: The property is inherent in every model. E.g. the finite state space of a finite state machine model. Static property: The property can be statically evaluated. E.g. the required memory of a finite state machine model. Dynamic property: The property can only be dynamically

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Dynamic property: The property can only be dynamically

  • evaluated. E.g. the required memory of a C program.

(src: A. Jantsch)

slide-7
SLIDE 7

7 ESII: Spec & Modeling PARTII

Need for heterogeneous modeling

A system consists of different parts. E.g. data flow and control flow dominated parts. Different objectives apply for different parts. E.g. the system and its environment. Different parts are developed by different people and tools.

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Different parts are developed by different people and tools. E.g. HW and SW.

(src: A. Jantsch)

slide-8
SLIDE 8

8 ESII: Spec & Modeling PARTII

So, what is a system?

Some definitions: “an aggregation or assemblage of things so combined by nature or man as to form an integral or complex whole” [Encyclopedia America] “a regularly interacting or independent group of items

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

“a regularly interacting or independent group of items forming a unified whole” [Webster’s Dictionary] “a combination of components that act together to perform a function not possible with any of the individual parts” [IEEE Standard Dictionary of Electrical and Electronic Terms]

(src: A. Jantsch)

slide-9
SLIDE 9

9 ESII: Spec & Modeling PARTII

Input-Output modeling process

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-10
SLIDE 10

10 ESII: Spec & Modeling PARTII

Mathematical models of real systems

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-11
SLIDE 11

11 ESII: Spec & Modeling PARTII

Example: temperature controller

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-12
SLIDE 12

12 ESII: Spec & Modeling PARTII

Static and dynamic systems

Definition: A static system is one where the

  • utput y(t) is independent of past values of

the input u(t0), t0 < t for all t

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Definition: A dynamic system is one where the output y(t) depends on the current input value u(t) and on at least another input value u(t0) with t0 < t, y(t) = f(u(t), u(t0)).

(src: A. Jantsch)

slide-13
SLIDE 13

13 ESII: Spec & Modeling PARTII

Time-varying and time-invariant systems

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-14
SLIDE 14

14 ESII: Spec & Modeling PARTII

Example: time-invariant system

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-15
SLIDE 15

15 ESII: Spec & Modeling PARTII

The concept of state

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-16
SLIDE 16

16 ESII: Spec & Modeling PARTII

State space

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-17
SLIDE 17

17 ESII: Spec & Modeling PARTII

State space model for continuous time, continuous state systems

(i.e. differential equation; g - state function)

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-18
SLIDE 18

18 ESII: Spec & Modeling PARTII

State space model for discrete time systems

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-19
SLIDE 19

19 ESII: Spec & Modeling PARTII

State space model time-invariant discrete time

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-20
SLIDE 20

20 ESII: Spec & Modeling PARTII

Linear and non-linear systems

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-21
SLIDE 21

21 ESII: Spec & Modeling PARTII

Deterministic, stochastic and non-deterministic systems

Definition: A system model is deterministic if the

  • utput function f and the state function g are

functions in the sense that they evaluate a given argument always and unambiguously to the same result.

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

result. A system model is stochastic if at least one of their

  • utput variables is a random variable.

A system model is nondeterministic if a given input may result in different outputs.

(src: A. Jantsch)

slide-22
SLIDE 22

22 ESII: Spec & Modeling PARTII

Events

Events are associated with a time instance and have no duration. Arrival of a message; Change of a signal value;

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Change of a signal value; Change of a state; A counter exceeding a given threshold value; An elapsed time period; etc.

(src: A. Jantsch)

slide-23
SLIDE 23

23 ESII: Spec & Modeling PARTII

Time-driven and event-driven systems

In time-driven systems the advance of time causes the system to become active. In event-driven systems the occurrence of an event causes the system to become

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

an event causes the system to become active.

(src: A. Jantsch)

slide-24
SLIDE 24

24 ESII: Spec & Modeling PARTII

Systems classification summary

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-25
SLIDE 25

25 ESII: Spec & Modeling PARTII

The Rugby Meta-Model

Abstraction in four domains Computation

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Communication Time Data

(src: A. Jantsch)

slide-26
SLIDE 26

26 ESII: Spec & Modeling PARTII

Example of a hierarchy

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-27
SLIDE 27

27 ESII: Spec & Modeling PARTII

Example of an abstraction

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-28
SLIDE 28

28 ESII: Spec & Modeling PARTII

Ways to handle complexity

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-29
SLIDE 29

29 ESII: Spec & Modeling PARTII

Hierarchy, Abstraction, Domain

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-30
SLIDE 30

30 ESII: Spec & Modeling PARTII

Rugby

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-31
SLIDE 31

31 ESII: Spec & Modeling PARTII

Domains in Rugby

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-32
SLIDE 32

32 ESII: Spec & Modeling PARTII

Computation Domain

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-33
SLIDE 33

33 ESII: Spec & Modeling PARTII

An MOS transistor model

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-34
SLIDE 34

34 ESII: Spec & Modeling PARTII

A transistor as switch

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-35
SLIDE 35

35 ESII: Spec & Modeling PARTII

An AND gate as transistor network

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

Two problems with arbitrary transistor networks:

Output is not defined when input is 0. Voltage drop between drain and source is relevant but not visible. that means, the model has not enough details in order to describe correctly the desired behavior

(src: A. Jantsch)

We can restrict the transistor network to a small number of patterns that Can be combined in arbitrary networks w/o violating assumption of switch-based transistor model

slide-36
SLIDE 36

36 ESII: Spec & Modeling PARTII

An Inverter as transistor network

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-37
SLIDE 37

37 ESII: Spec & Modeling PARTII

Gate-based abstraction level

  • 1. The primitive elements are defined by simple

models, i.e. small truth tables in this case.

  • 2. The primitive elements can be implemented in a

wide range of technologies.

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

wide range of technologies.

  • 3. The model holds even for arbitrarily large

networks of primitive elements.

(src: A. Jantsch)

slide-38
SLIDE 38

38 ESII: Spec & Modeling PARTII

Algorithms, Functions, Relations

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-39
SLIDE 39

39 ESII: Spec & Modeling PARTII

Sorting defined as a relation

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-40
SLIDE 40

40 ESII: Spec & Modeling PARTII

Sorting defined as a function

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-41
SLIDE 41

41 ESII: Spec & Modeling PARTII

Sorting defined as algorithms

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-42
SLIDE 42

42 ESII: Spec & Modeling PARTII

The communication domain

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-43
SLIDE 43

43 ESII: Spec & Modeling PARTII

Communication at layout and gate level

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-44
SLIDE 44

44 ESII: Spec & Modeling PARTII

Communication between processes

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-45
SLIDE 45

45 ESII: Spec & Modeling PARTII

The data domain

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-46
SLIDE 46

46 ESII: Spec & Modeling PARTII

The time domain

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-47
SLIDE 47

47 ESII: Spec & Modeling PARTII

Notation for abstraction levels

computation communication

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de

data time

(src: A. Jantsch)

slide-48
SLIDE 48

48 ESII: Spec & Modeling PARTII

Abstraction levels in design phases

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

design phases domains

slide-49
SLIDE 49

49 ESII: Spec & Modeling PARTII

Design activities in terms of abstraction levels

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

Note: Design activities typically make design decisions and thus refine the design into a model at a lower abstraction level But: analysis activities do not refine models

slide-50
SLIDE 50

50 ESII: Spec & Modeling PARTII

Analysis activities in terms of abstraction levels

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

Analysis: check for consistency,; produce estimates;

  • In early design phases -> check for given requirements, for mutual

consistency or infer more detailed constraints

  • May also check feasibility
slide-51
SLIDE 51

51 ESII: Spec & Modeling PARTII

The network terminal case study

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

CPN: customer premises network

slide-52
SLIDE 52

52 ESII: Spec & Modeling PARTII

The network terminal case study (cont’d)

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

  • Traffic model: based on Asynchronous Transfer Mode (ATM)
  • ATM: virtual channels can share the same physical transport

Medium (like fiber)

  • ATM switch routes data packets from different virtual channels in the access network to the

respective CPN interface and vice versa

slide-53
SLIDE 53

53 ESII: Spec & Modeling PARTII

The NT design flow

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-54
SLIDE 54

54 ESII: Spec & Modeling PARTII

Models in the NT design

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-55
SLIDE 55

55 ESII: Spec & Modeling PARTII

The network terminal case study (cont’d)

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-56
SLIDE 56

56 ESII: Spec & Modeling PARTII

Transformations in the NT design

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de (src: A. Jantsch)

slide-57
SLIDE 57

57 ESII: Spec & Modeling PARTII

References and Sources

[A. Jantsch] A. Jantsch, “Modeling Embedded Systems and SoCs”, Morgan Kaufmann Publishers, 2004.

  • J. Henkel, Univ. of Karlsruhe, WS0708

http://ces.univ-karlsruhe.de