Software Engineering For RealTime: A Roadmap

1
Software Engineering For Real-Time A Roadmap

• Author Hermann Kopetz
• Technical University of Vienna, Austria
• lthk_at_vmars.tuwien.ac.atgt
• Presenter George Edwards
• University of Southern California
• ltgedwards_at_usc.edugt

2
Presentation Overview

• Introduction
• Technology Trends
• Requirements for Real-Time Systems
• Composability
• Validation
• Critique of the Paper

3
Introduction

• Increasingly widespread and important
• Avionics, telecommunications, medicine, and
  defense applications
• New challenges for software engineers
• Resource constraints
• Stringent quality-of-service (QoS) demands
• Availability, security, scalability
• Composability of COTS components
• Validation of component-based architectures

Distributed, Real-Time, Embedded (DRE) Systems
4
Technology Trends (1/2)

• Advances in the semiconductor industry will have
  a strong influence on SE for RT
• RT systems composed of two types of hardware
  components
• System-on-a-chip (SOC)
• High dependability, low-production costs
• Smart microelectronic mechanical systems (MEMS)
  sensors
• Signal conditioning, calibration, diagnostics,
  and network control

5
Technology Trends (2/2)

• Economic factors demand that complex, large-scale
  systems be composed of commercial-of-the-shelf
  (COTS) components
• A component is a self-contained subsystem used as
  a building block in a larger system
• Hardware components
• Software components
• Combined hardware/software components
• A self-contained SOC with OS and RT application

6
SE Requirements for DRE Systems (1/3)

• Two-level design methodology
• Strict separation of architectural design and
  component development
• Architectural level component interactions and
  communication network interfaces
• Component level implementation of defined
  services
• Requires precise specification of CNIs in both
  value and temporal domains
• Enables system analysis without knowing component
  implementations

7
SE Requirements for DRE Systems (2/3)

• Predictable communication
• Two types of real-time networks
• System networks connect SOCs
• Sensor networks connect SOCs to smart sensors
• Both network types must
• Have stable temporal properties
• Provide a global time of known precision

8
SE Requirements for DRE Systems (3/3)

• Generic fault-tolerance
• Architectures that provide generic services
• Achieved through distributed communication
  control and replication
• Provided by the network, hardware, and system
  software
• Application software for fault-tolerant and
  non-fault-tolerant systems is the same

9
Composability

• Components must be autonomous and encapsulated
• Various possible vantage points
• Service provision
• Validation
• Error containment
• Reuse
• Design and maintenance
• Design methodology must ensure these properties
  are not invalidated by system integration.

10
Service Classes

• Analysis of component-based architectures should
  distinguish
• Prior services, provided by an individual
  component
• Emerging services, provided by integration of
  multiple components

11
Component Interfaces

• Real-time Service (RS) interface
• Hides component implementation
• Provides a service according to the CNI spec
• Diagnostic and Management (DM) interface
• Exposes component implementation
• Allows configuration and fault diagnosis
• Configuration Planning (CP) interface
• Generates glue code

12
Principles of Composability

• Independent development of components
• System architecture does not rely on specific
  component implementations, only interfaces
• Stability of prior services
• Validated services not affected by integration
• Constructive integration
• Introduction of a new component does not
  interfere with correct system operation

13
Validation

• Composable architectures will shift emphasis back
  to product validation
• Component designers will need to establish tight
  upper bounds for all time-critical processes
• Rare-event simulations (faults, peak-load
  performance) will be required
• Formal verification methods will grow in
  importance

14
Critique of the Paper

• Strengths
• In depth analysis of composability and validation
• Very specific view of the future of RT systems
• Weaknesses
• Somewhat narrow focus ignores many important
  issues
• As much about hardware as software
• Relevance to embedded systems
• Embedded systems are almost always RT systems