Concurrent Engineering

1
Hardware/Software Codesign of Embedded Systems
CONCURRENT ENGINEERING
Voicu Groza SITE Hall, Room 5017 562 5800 ext.
2159 Groza_at_SITE.uOttawa.ca
2
CONCURRENT ENGINEERING

• General Issues
• System Development Life Cycle
• System Categories
• Reactive Real-Time Embedded Systems

3
Motivation for Concurrent Engineering

• Project size 5 500 people
• Time money
• Development time versus development cost
• Average product life time was 23 years
• For mobile phones it is 6 to 12 months
• 60 of Ericssons profit comes from products
  younger than 3 years
• Late arrivers loose market share
• Phase and activity overlapping is crucial
• Communication and documentation are essential
• The development task must be partitioned into
  horizontal and vertical subtasks.
• The development time must be shortened while
• - the design complexity increases and
• - the product quality level remains high.

4
Definitions and Scope

• CE Paralleling of life cycle functions
  Consensus Cooperation
• CE Minimization of total product development
  time
• CE Global optimization of total product live
  cycle (maximize quality, reduce lead time, lower
  cost)
• CE Integrate product and process design over
  the entire enterprise

5
Influencing Agents the 7 Ts
Tasks
Talents
Teamwork
CE
CE
CE
CE
CE
CE
Techniques
Tools
CE
Time
Technology
6
Project Phases
Expression of need
Project definition
Abstraction

• Project management

Planning organization
Project development
Project completion
Time
7
Involved People

• Customers
• - Users
• - Marketing and sales personnel
• - Operators
• - Maintenance personnel
• Product manager
• Project manager
• Requirement definition engineer
• Specification engineer
• Designer
• Implementation engineer
• Test engineer

8
Development Phases

• Requirement definition The most informal part
  of a product development
• Specification The first abstract description
  of the product
• Design Functional partitioning and
  architecture selection
• Implementation Implementation of components
  and integration
• System validation
• Production Experimentation with prototypes and
  final product production
• Operation Corrective, adaptive and preventive
  maintenance

9
Life Cycle Models

• Waterfall model
• V-cycle
• Spiral model
• Contractual model

10
The Waterfall Model
Requirements Definition
Specification
Design

• Based on the assumption of document completion at
  the end of each stage. This is problematic for
  applications for which the requirements and
  implementation are poorly understood.

Implementation
Testing Maintenance
11
The V- Cycle Model
NEED
PRODUCT
Certification
Validation
Validation
DESIGN
VALIDATION
12
The Spiral Model - Risk Driven
CUMULATIVE COST
Determination of objectives, alternatives and
constraints
Evaluate alternatives, identify resolve risks
Risk analysis
operational prototype
Risk analysis
prototype
RA
prototype
Life cycle plan
SW product design
concept
Detailed design
Development plan
requirement
Unit test
Design validation verification
Integration test
Integration test
Develop Verify
Planning next phases
Implementation
13
Concurrency Between Process Phases

• Align each step as far to the left as possible
  
• Maintain the precedence of tasks
• Minimize the horizontal overlap between two
  consecutive tasks
• Maximize the independence of tasks

Product
14
Design Phase OverlapWaterfall Model
TIME
15
Design Phase OverlapConcurrent Engineering
EFFORT
Test
Implementation
Design
Specification
TIME
16
Effort Distribution
17
Error Correction Cost
COST
10000
Log scale
1000
100
10
1
0.1
Time of Error Detection
Specification Design Implementation Production
Operation
18
Time to Market Cost Model
Revenue /mo
Market rise
Market fall
Time to market
Time
System concept
Production
19
Productivity Factors
experience with design language
experience with tools
development of tools
design tools
high-level modeling
experience with the application
real time constraints
product complexity
team ability
Productivity
20
Basic Principles of CE

• Early problem discovery Problems discovered
  early are easier to solve.
• Early decision making In the beginning the
  design space and the potential to find global
  optima is greater.
• Work structuring Humans are not good on
  working on multiple tasks simultaneously they
  are good on systematically structuring their work
  and the work environment so that each task is
  independent.
• Teamwork affinity Cooperation between persons
  and teams is based on trust and positive
  experience.
• Knowledge leveraging Since the required
  knowledge is distributed, the decision making
  process must coordinate distributed knowledge.
• Ownership The motivation is higher if teams or
  persons own a problem and assume responsibility.
• Consistent objectives Persons, groups, and
  organizations have self interests which might not
  be consistent with the companies objectives.

21
Concurrency and Simultaneity

• Parallel product
• Concurrent resource scheduling
• Concurrent processing
• Minimize interfaces
• - Minimize product interfaces
• To facilitate handling, maintenance, testing, and
  manufacturing
• To facilitate concurrent activities
• - Minimize process interfaces
• - Automate data flow
• Efficient communication and efficient reviews
• Design documentation and design decision
  documentation
• Quick processing of individual activities

22
Partitioning of Development Tasks Vertical
Slices
23
Horizontal Partitioning Structural and
Functional Slices
STRUCTURAL COMPONENTS
Implementation
Specification
Integration
Validation
Design

• The number of interdependent tasks must be
  minimized

24
Functional Slicing

• Functional slices may or may not be implemented
  by structural components
• Functional slicing increases the potential for
  concurrent activities
• Functional slicing allows to start with system
  simulation and system validation very early.
• Early system level validation generates bug
  reports and valuable feed back for the design of
  individual components.

25
Benefits of Concurrent Engineering

• Rationalization in the manufacturing process
• Working parallel
• Improved communication and providing better
  input
• Preempting errors and spotting problems early
• Flexibility to accommodate changes
• Decreased occurrence of obsolescence
• Cross training
• Better use of scarce technical resources

26
Pitfalls

• Problem with handling of information and data
  flow
• Risk of wasted efforts due to parallel
  activities
• Error correction activities due to incomplete
  and immature information
• Increased development cost

27
System Categories

• Embedded systems
• Real-time systems
• Distributed systems
• Interactive systems
• Reactive systems
• Communication systems (protocol processing,
  switches, etc.)
• Data processing systems (processing of signals,
  pictures, voice, etc.)

28
Embedded Systems

• Embedded electronic systems implement
  application specific dedicated functions in a
  special environment.
• Dedicated systems
• Environment is complex
• Environment establishes various nonfunctional
  constraints
• Environment is not completely known
• Interfaces to other subsystems
• - Digital protocols
• - Analog interface
• - Mechanic interface
• - Chemical interface
• Sensors and actuators

29
Real-time Systems

• Real-time systems have a well defined timing
  behaviour imposed upon them by the environment.
• Well defined timing behaviour
• Timing constraints
• - Local timing constraints
• - Input rate constraints
• - Output rate constraints
• Several simultaneous activities
• - Observe events
• - Evaluate decisions
• - Generate actions
• Synchronization facilities

30
Distributed Systems

• Distributed systems implement an integrated
  functionality at spatially separated locations.
• Communication facilities and protocols
• - Specification. modeling, implementation, and
  verification of communicating systems
• - Synchronous and asynchronous communication
• - Shared memory
• - Shared bus
• - Point-to-point protocols and structures
• Distributed control threads
• - Parallel threads
• - Dynamic creation and destruction of threads
• - Remote procedure call (synchronous and
  asynchronous)
• High complexity

31
Interactive Systems

• Interactive systems have an intense and versatile
  data flow to I/O devices with soft timing
  constraints on performance and response time.
• Complex mixed HW/SW user interface
• Soft timing constraints which must be met on
  average

32
Reactive Systems

• Reactive systems are connected to the environment
  via sensors and actuators and implement control
  and data processing functionality.
• Versatile interface and communication
  facilities and protocols
• - Interfaces to mechanical and analog devices
• - Many nonfunctional constraints (size, power,
  shape, weight, etc.)
• Hard timing constraints on response time
• The behaviour is well modeled by a finite state
  machine

33
Communication Systems

• Communication systems process and connect
  streams of data according to protocols and
  switching rules.
• Tight performance constraints on throughput
• Moderate timing constraints on latency
• Large memory requirements with irregular access
  pattern
• Large number of internal interconnections and
  I/O ports
• Implementation of the multi-layer protocol
  stack is partly in HW and partly in SW

34
Data Processing Systems

• Data processing systems transform streams of data
  according to complex mathematical transformations
  with high performance requirements.
• Tight performance constraints on throughput
• Varying memory requirements with regular access
  pattern
• Only one or few data streams
• Complex mathematical transformations
• Implementation is partly in custom HW and
  partly in SW on DSPs

35
Development Context
Activities

• Company

Company management
Projects
Project management
Project development
Resources