CS4730 Real-Time Systems and Modeling

1. CS4730Real-Time Systems and Modeling Fall 2010 José M. Garrido Department of Computer Science & Information Systems Kennesaw State University

2. Real-Time Systems RTS are computer systems that • monitor, • respond to, or • control an external environment (C) J M Garrido

3. Environment The environment is connected to the computer system through: • Sensors • Actuators • Other I/O devices (C) J M Garrido

4. General Architecture of A Real-Time System (C) J M Garrido

5. Real-Time System (C) J M Garrido

6. Constraints • A real-time system must meet various timing and other constraints imposed by the real-time behavior of the environment • A real-time system is also called a reactive system because it must respond to or react to signals from the environment (C) J M Garrido

7. Real-Time Systems • A reactive system that maintains an on-going interaction with its environment, and with given time windows for its output • Includes a combination of hardware and software • There is a wide variety of RTS • small embedded systems • very large systems on WANs. (C) J M Garrido

8. Examples of RTS • Vehicle systems for automobile, subways, aircraft, railways, ships, … • Traffic control • Process control for power plants, chemical plants, … • Medical systems • Military applications • Manufacturing systems with robots (C) J M Garrido

9. Examples of RTS (Cont.) • Communication systems • Computer games • Multimedia systems that provide text, graphic, audio, and video interfaces • Consumer appliances (C) J M Garrido

10. Embedded Systems • A real-time system that is part of another larger system • Does not usually interact with humans users or operators • Interfaces directly with sensors and actuators • Very specialized to suit a specific purpose • The software is usually implemented in firmware (C) J M Garrido

11. Embedded Real Time System (C) J M Garrido

12. Categories of Real_time Systems Different systems will have different requirements for meeting deadlines: • Hard real-time systems - missing even a single deadline is considered unacceptable, e.g., The new U.S. Air Traffic Control System. • Soft real-time systems - missing a deadline occasionally is considered acceptable, e.g., telephony systems. (C) J M Garrido

13. Real-Time Systems General characteristics: • Reactive, the system must respond timely to internal and external input events • Concurrent, the system needs to manage several tasks (or processes) • Each task has timing constraints, i.e., it has to respond in a timely fashion. (C) J M Garrido

14. Other Major Characteristics • Reliability - a measure of how often a system will fail. Also stated as the probability that a system will perform correctly for a given period • Fault tolerance - recognition and handling of failures. Avoidance of failures is important, but responding appropriately to failures is challenging (C) J M Garrido

15. Criticality • A measure of the failure cost. The higher the cost of failure, the more critical the system. • A critical system might control an aircraft or a nuclear power plant • A RTS might be a hard system, even though it is not a critical one. For example, a computer game. (C) J M Garrido

16. Architecture of Real-Time Systems • Hardware components • used to interact with the environment • consists of sensors and actuators • Software components • controls the actions of the hardware • computations (C) J M Garrido

17. Properties of Real-Time Systems • Timeliness - the system must perform operations in timely manner • Reactiveness - the system continuously responds to (random) events • Concurrency - multiple simultaneous activities are carried out • Distribution - tasks cooperate in multiple computing sites (C) J M Garrido

18. Timeliness Issues • The goal is to reduce two specific intervals: • service time - the interval taken to compute a response to a given input • latency - the interval between the time of occurrence of an input and the time at which it starts being serviced • The sum of these two intervals represents the response time. This must be shorter than the deadline for this type of input. (C) J M Garrido

19. Safety Properties • The system should never be in certain (undesirable) states • “The system should not do bad things” • In the Train-gate System, the gate should never be open when a train is in the intersection area (C) J M Garrido

20. Progress or Liveness Properties • These describe the required “live” and desirable behaviors of a system • For example, in the Train-Gate System, the gate should be open whenever there are no trains in the area (C) J M Garrido

21. Techniques for Studying RT Systems The techniques for dealing with timeliness problems belong to two main groups: • Analytic techniques - use formal mathematical models of the systems and mathematical techniques • Simulation - use computer models to compute behavior characteristics. (C) J M Garrido

22. Basic Elements in Behavior • Events, are instantaneous occurrences that trigger transitions and operations on some objects. Events can be external or internal. • States, represent a unique behavior with respect to the other states in an object. An event or message can cause a change of state. • Transitions are state changes in the objects. (C) J M Garrido

23. Process and States • In a high-level view, a RTS is considered a closed world containing two components: an environment and a computer system • These components interact via messages, signals, or events at their interface • At any time, the system may be in one of several states (C) J M Garrido

24. States • Each state denotes a particular set of values of variables or set of relations among variables • The system changes state due to events that change the values of, or relations among, these state variables • These state changes are called transitions (C) J M Garrido

25. Model of RTS • The software that implement RT systems deal with states, and transitions and a set of interacting processes • Processes interact by • sending messages and signals to each other • sharing hardware and software • competing for resources • A process is an active component (object) of a RTS (C) J M Garrido

26. Reactiveness Issues • Transformational systems - some initial data is transformed by a series of computations into the desired output data, then the system terminates • Reactive systems - are involved in a continuous interaction with the environment, these systems do not terminate. (C) J M Garrido

27. Nondeterminism • In general, real-time systems are non-deterministic systems since they have no control over the relative order or time of occurrence of input events. • This characteristic is a reflection of the unpredictable nature of the real-world in which such systems operate. (C) J M Garrido

28. Concurrency Issues • A task is a sequential thread of control, a system with concurrency contains two or more simultaneous threads of control that dynamically interact • Threads may progress at different speeds, their interactions can be: • synchronization • communication (C) J M Garrido

29. Communication The communication of tasks in real-time systems can be: • Synchronous - a direct communication between two tasks. Both the sender task and the receiver task have to be ready • Asynchronous - an indirect communication that uses mailboxes. UML Sequence diagrams are used to describe this. (C) J M Garrido

30. Distribution Issues • Consists of a set of computing sites loosely coupled through a communication network • Message loss is just one of a collection of truly difficult problems facing designers of distribution systems (C) J M Garrido

31. Active and Passive Agents • Active agents are the major components of a system with their own thread of control • Passive agents only “executes” when one of its operations is invoked - it is only active during interactions with other objects • Design and programming with agents involves constructing “machines” and interconnecting them. (C) J M Garrido

32. Types of Processes in RTS • Periodic processes - are activated in a regular basis between fixed time intervals. Used for systematically monitoring, polling, or sampling data from sensors • Sporadic processes - are activated by an external signal, i.e., are event-driven. (C) J M Garrido

33. Periodic Process • A periodic process is characterized by a triple (c,p,d). • The computation period is denoted by c • The period or cycle time is denoted by p • The deadline is denoted by d • The first important relationship is: c  d  p. (C) J M Garrido

34. Sporadic Process • A sporadic process is also characterized by a triple (c,p,d), with c  d  p. • On the occurrence of the event, the computation must be completed within the specified deadline. The completion time t must satisfy: t  te + d te is the event occurrence time (C) J M Garrido

35. Further Timing Effects • The interactions among processes further add more delays in the computation times of the processes • Another effect is jitter. This is the variation from cycle to cycle, or from event to event, of the time to complete a task. (C) J M Garrido

36. Performance Measures for RTS • Latency • Number of deadlines missed • Worst case reaction time (C) J M Garrido

37. General RTS Approach (C) J M Garrido