Strategic Directions in RealTime and Embedded Systems

1
Strategic Directions inReal-Time and Embedded
Systems

• CSCI 589 presentation
• September 14, 2004
• Kyuhyung Lee

2
Before beginning

• How to approach
• not what, but how
• careful of duplicated examples and overlapping
• trace the relationship topics and examples
• Keyword
• real-time
• predictability
• open system
• QoS guarantee

3
Situation

• Changes in application areas
• time critical
• larger and more complex
• Embedded systems
• pervasive but independent
• not much consideration of how and when
• What do we need ?
• predictability for the result
• more consideration of economic and safety

4
Overcome

• Real-time systems
• not only the logical result,
• but also the time at which the results are
  produced
• component of larger systems
• not good enough yet, still need to be
  developed
• Strategic directions in development and research
• with the systems
• in the related areas
• 
  

5
Success so far
Validation technology
Software industry still growing
Validation technology
Software industry
schedulability conditions and efficient, robust
and accurate validation algorithm
avionics, medical, communication, consumer and
instrumentation
QoS guarantee
Precise response
R/T operating systems
Real-time computing
6
Challenges

• High-level challenges
• system evolution
• open real-time system
• composibility
• software engineering
• Issues in related areas
• science of performance guarantee
• reliability and formal verification
• general system issues
• real-time multimedia
• programming languages
• education

7
Challenges - continue

• System evolution
• Open systems and composibility no more new
  installation
• use of open-standard-based component
• coherent set of interfaces
• convenient and safe environment
• For off-the-self component
• Scheduling and resource management schemes
• predictability covering functionality,
  timeliness
• and fault-tolerance
• Schemes to compose the subsystems into larger
  systems

8
Challenges - continue

• Open real-time systems
• Not only achieving a specific set of goals
• purchase and run multiple application with
  real-time requirements
• New real-time system architecture
• general-purpose
• co-existence of multiple and independently
  developed applications
• Difficulties of open real-time computing
• hardware characteristic
• aggregate resource and timing requirements
• perfect a priori schedulability analysis
• Economic and correctness criteria

9
Challenges - continue

• Composibility
• Lagacy systems
• components focusing on functional composition
• to be evolved to be robust while delivering high
  real-time performance
• New way of composition
• Interacting domains function, time and fault
  tolerance
• lead to adaptive high-performance fault-tolerant
  embedded systems
• need new environments of programming and
  operating
• New environments
• strong analysis tool with respect to meeting
• fault-tolerance and real-time constraint
• O/S with dynamic, flexible and adaptive behavior
• under time constraints, easy interoperability
  and easy porting

10
Challenges - continue

• Software engineering
• In these days
• motivated by large-scale complex systems
• research and product only focusing on functional
  issues
• nonfunctional issues has been left to special
  version
• -gt time and labor consuming work CORBA
• New environment to undergo
• time, dependability, and other QoS constraints
• evolvability separating platform-dependent
  concerns from applications
• structure for composible module not only
  functionality
• adaptable and configurable design
• derive and impose timing constraints

11
Challenges - continue

• Science of performance guarantees
• Classical real-time systems
• off-line deterministic guarantees in meeting
  safety and real-time requirements subject to
  failure and environmental assumptions
• larger, more dynamic and deployed in
  nondeterministic and less safety-critical
  environments
• New science of performance guarantees
• more formal analysis of dynamic real-time systems
• need for probabilistic guarantees new QoS
  requirements
• schedulability condition and validation algorithm

12
Challenges - continue

• Science of performance guarantees - continue
• Schedulability condition and validation algorithm
• based on workload model
• measured in terms of complexity, robustness and
  degree of success
• can reduce the need of accurate characterization
  of the applications and run-time environments
• timing validation
• Limitations of current schedulability condition
  and validation algorithm
• low degrees of success with sporadic processing
  requirements
• neither applicable nor robust with probabilistic
  timing constraints
• Only for statically configured systems

13
Challenges - continue

• Reliability and formal verification
• Reliability
• computers become components of complex systems
• dynamic analysis based on testing and run-time
  monitoring and checking
• common framework of analysis for large-scale
  system
• Common framework
• from current formalisms
• logics, automata and state machines, Petri nets
  and process algebras
• test scheduling logics and process algebras
  based on specifications
• Much more issues from implementation
• traceability from specification to low-level
  implementation
• multiple specifications in terms of state machine
  and Petri nets
• - monitoring and checking at run time

14
Challenges - continue

• General system issues
• Not that serious but basic
• How to do at worst-case execution time ?
• architecture and various cache
• How to implement and analysis ? - predictability
• How to validate ? real-time and resource
  management workload model
• How to develop ? design, environment support
• And so on transaction guarantee and real-time
  database

15
Challenges - continue

• Real-time multimedia
• Multimedia computing and communication
• Humans interacting with information in
  probabilistic nature
• Issues
• precise specification predictability
  requirements
• Algorithm to map the predictability requirements
  to the mechanism
• system-level predictability properties,
• if individual and absolute guarantees exist
• QoS guarantees for composing and integrating
• New QoS guarantee
• Using queuing theory focus on equilibrium and
  aggregation
• Real-time scheduling theory
• Behavior associated with queuing theory

16
Challenges - continue

• Programming languages
• Weakness
• lack of temporal and real-time requirements
• not support many aspects of guarantees and
  scheduling schemes
• New paradigm
• more effective structuring mechanisms
• consider nonfunctional aspects
• OOP paradigm
• wide range of requirements, timing guarantees and
  scheduling

17
Challenges - continue

• Education
• Teaching real-time systems for students
• manipulate programs and systems
• mathematically and component based within
  larger systems
• get knowledge of interacting with processes in
  the real world
• Category
• execution of a sequential program in time
• - how to handle real-time constraints
• management of logical and physical concurrency in
  time
• - cooperating sequential processes
• - competition for shared resources among
  processes

18
Vision

• Next decade
• significant fraction of the global population
• highly distributed environment
• demand for safe, dependable and certifiable
  real-time systems
• Key roles of real-time systems
• High-integrity real-time services delivered on
  time
• QoS attributes
• High-level temporal control regarding the
  ubiquitous nature
• open systems with nonfunctional issues

19
Critique

• Strength of approach
• step by step opinion, examples and vision
• common and easy examples
• Weakness of approach
• ambiguous and duplicated explanation
• distributed topics hard to find their
  relationship
• Relevance to embedded systems
• time and validation issues
• nonfunctional factor dependability, QoS
  guarantee,
• availability of resources and so on
• more close characteristic to embedded
  systems

20
Q A

• Whatever