Software Engineering for RealTime: A Roadmap By Hermann Kopetz

1
Software Engineering for Real-TimeA Roadmap-
By Hermann Kopetz

• Prakash Gupta
• Sept 28th 2007
• prakashg_at_usc.edu

2
Overview

• Introduction
• Soft vs hard real-time systems
• Technology trends
• What is required for DRE?
• Composability
• Validation
• Recent challenges
• Applications

3
Where?
Avionics
4
Where?
Automotives
5
Where?
Medicine
Ultrasound imaging
Dialysis machine
6
Soft vs Hard real-time systems

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Soft real-time
• Statistical margin of error
• No significant financial loss

• Hard real-time

• Hard real-time
• Mission critical systems
• Catastrophic consequences

7
Soft vs Hard real-time systems

• Hard real-time
• Temporal domain is as critical as value domain
• Soft real-time
• Temporal domain is not critical as value domain

Design of HRT is fundamentally different than
that of SRT
8
Real-time task models
Real-time task models

• Periodic
• Continuous deterministic pattern of time
  interval
• Characterized as a tuple (C,T)
• Eg. robotics application
• Sensor data network transmission

9
Real-time task models

• Aperiodic
• Non-deterministic request periods
• Event driven real-time systems
• Eg. Ejection of a pilot seat

10
Technology trends

• System on a chip
• Integrating all components on a single chip
• Cost-effective if mass-produced

11
Technology trends

• Smart MEMS Sensors
• Sensor with a microcontroller
• Advantages
• Weak non-linear sensor signals
• Local monitoring
• plug-and-play capability
• Complexity abstraction

12
What is required for DRE?

• Two-level design methodology
• Design of architecture Clear separation
• Development of components
• Enables system analysis without knowing component
  implementation

13
What is required for DRE?

• Predictable communication
• QoS should be predictable
• Real-time network types
• System network
• Distributed communication protocol
• Fault-tolerance
• Sensor network
• Multi-master networks
• Fault-tolerance

14
What is required for DRE?

• Generic fault-tolerance
• Architectures that provide generic services
• Fault-tolerant clock synchronization
• Membership service at H/W or system S/W level

15
Composability

• Properties at component level must hold at system
  level
• Properties of an ideal component
• Service provision
• Validation
• Error containment
• Reusability
• Design and maintenance

Grand Challenge!!! Architecture for distributed
real-time systems that support the principle of
COMPOSABILITY
16
Principles of Composability

• Independent development of components
• Precise CNI specification in value temporal
  domain
• Stability of prior services
• Ensures validated service of component is not
  refuted while integration
• Constructive integration
• Design of communication system

17
Validation

• Composable architectures will shift focus back to
  product validation
• Knowledge about WCET
• Rare event simulations
• Validate fault-tolerance
• Peak-load performance
• Formal verification
• Critical algorithms

18
Recent Challenges

• Optimized use of system resources at design phase
• Open loop approach vs Feedback approach
• Consistency synchronization
• Specialized fault-tolerance and security

19
Brake Pads

• Dynamically measure the pressure
• TactArray Sensors pressures up to 2000 psi at
  temperatures up 200C

20
Gastrointestinal Diagnostic equipment

• Measures the pressures applied by muscles in the
  GI tract

21
FingerTPS

• To teach doctors performing physical
  manipulations in a consistent and repeatable way
• System records and displays finger and palm
  pressures exerted during treatment

22
Critique of the paper

• Strengths
• In depth analysis of composability
• Good insight into the future of RT systems
• Drawbacks
• Overlooked most of the challenges

Most of the time embedded systems are real-time
systems
23
References

• Open Challenges in Real Time Embedded Systems
  Lui Sha, CS, UIUC
• http//www.pressureprofile.com
• http//www.real-time-systems.com
• http//cs-people.bu.edu/gfry/realtime.html
• http//en.wikipedia.org

24

• Comments
• ?