© Adeel Pasha, LUMS, Spring 2018

EE-522: Lecture-2

Basics of Real-Time Scheduling

 © Adeel Pasha, LUMS, Spring 2017

Recap

Safety and Reliability

2 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Safety and Reliability: A Closer Look

 Non-RT systems: Safety and Reliability are independent
 A word processing software (might be unreliable but safe)
 A handgun (might be reliable but unsafe)
 Fail-Safe State
 If system enters this state upon failure, no damage would be done
 Every non-RT system has ≥1 fail-safe state(s)
 It makes an unreliable & unsafe system into a safe & reliable system
 Examples:
 MS Word when crashes, saves the last modifications
 Traffic signals
 Fail-safe state??
 Every RT-system does not necessarily have a fail-safe state
 Failure of the system can cause severe damage  safety-critical

3 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Safety and Reliability: A Closer Look

 Safety-critical
 Safety can only be ensured through increased reliability
 Standard reliability criterion for a flight-control computer?
 1 failure per 109 (1 Giga) flying hours
 How to achieve Reliability?
 Error avoidance (formal verification) C1
V
 Error detection and correction (EDAC) O
 Fault-tolerance C2
T
I
 Built-In Self Test (BIST) N
 Triple Modular Redundancy (TMR) G
C3

4 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Safety and Reliability: A Closer Look

 Software Fault-Tolerance Techniques
 N-Version Programming
 Similar to TMR for HW faults
 Have N different versions of a program developed by different teams
 @ runtime, these N versions give results and majority voting is used
 There could still be failures due to statistical co-relation
 If an exam question is difficult for you, it would be difficult for
everybody (conditional statement btw !)
 Recovery Blocks
 Check-pointing and Roll-back

5 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Safety and Reliability: A Closer Look

Recovery Blocks

Source: Real-time Systems: Theory and Practice by Rajib Mall

6 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2017

Taxonomy of Real-Time Tasks

7 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Types of RT Tasks 100%

Deadline

Utility

 Hard RT task 0
Response Time

 Considered failed if deadline is surpassed

 Collision detection & avoidance tasks for a robot

 Firm RT task
 There is a defined deadline
 But failure to meet deadline doesn’t mean system failure
 Live video streaming
 Temperature monitoring using sensor networks
8 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Types of RT Tasks 100%

Deadline

Utility
 Soft RT task 0
Response Time
 The timing constraints are not absolute
 Average response time is measured
 Internet connectivity (in general), web browsing

 Non-RT task
 Any example!

9 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Events in RT-System
 Stimulus Events
 Generated by environment and used by system
 Periodic, Aperiodic

 Response Events
 Produced by system and used by environment

 Example: dial-tone reception on lifting the receiver

 Stimulus: detaching the hand-set receiver
 Response: dial-tone generation
10 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Types of Timing Constraints

 Delay constraint
Δ ≥ delay
 Minimum time (delay) must pass
between two events
t(e1) delay t(e2)
t =0
 Deadline constraint
 Maximum permissible separation
between two events should not Δ ≤ deadline
be surpassed
t(e1) t(e2)
deadline
 Duration constraint t =0
 An event must occur for a certain
duration of time
11 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Modeling Timing Constraints: EFSM*

 Case Study of a Land-Line Telephone System
Stimulus Event

Person Telephone
(Environment) (System)

Response Event
* EFSM: Extended Finite State Machine
12 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Modeling Timing Constraints

 Deadline Constraints
 Stimulus-Response (S-R) Type
 Also called Performance Constraint Await
1st Digit
Await
Dial-Tone/
Dial-
Set Timer (30s)
Tone

Timer
Await
Receiver Expired/
Receiver
Lifted/ Beeping
Replacement
13 EE-522: Embedded Systems
Set Timer (3s) 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Modeling Timing Constraints

 Deadline Constraints
 Response-Stimulus (R-S) Type
Await
 Also called Behavioral Constraint
2nd Digit

Await 1st Digit/

1st Digit Set Timer (3s)

Timer Await
Dial-Tone/ Expired/ Receiver
Set Timer (30s) Beeping Replacement
14 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Modeling Timing Constraints

 Deadline Constraints
 Stimulus-Stimulus (S-S) Type
Await 3rd
Digit
2nd Digit/
Await
Set Timer (3s)
2nd Digit

Timer Expired/
Await
1stDigit/ Beeping
Receiver
Set Timer (3s) Replacement
15 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Modeling Timing Constraints

 Deadline Constraints
 Response-Response (R-R) Type Await
Call
Pick-UP
Await
Ring-Back/
Ring-Back
Set Timer (1 min)
Tone

Timer
Await
Ring-Tone to Expired/
Receiver
Callee/ Beeping
Replacement
Set Timer (3s)
16 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Modeling Timing Constraints

 Delay Constraints
 Stimulus-Stimulus (S-S) Type Await 3rd
Timer Expired Digit
Await
2nd
Digit

2nd Digit/Beeping
Await
1st Digit/ Receiver
Set Timer (50ms) Replacement

17 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Modeling Timing Constraints

Timing
Constraints

Performance Behavioral

Delay/Deadline Duration Delay/Deadline Duration

S-R R-R R-S S-S

18 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2017

Basics of Real-Time Task Scheduling

19 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Important Notions
 Task Instance
 Every stimulus event triggers a response event
 A piece of code has to be run, etc.
 Could be periodic or aperiodic
 Ti(j)  Jth instance of Task i

 Most of the tasks are deadline-constrained tasks

 Relative vs. Absolute Deadline

 Absolute: Measured with the help of a real clock
 Relative: referenced to some other event
20 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Important Notions
 Relative vs. Absolute Deadline

Relative Deadline of Ti = di

Absolute Deadline of Ti (1) = φi + di

φi
Ti (1) Ti (2)

0
Arrival or activation of Ti (1)
21 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2018

Important Notions
 Task Precedence
 If a task must be completed before another task to start
 It’s recurring
 Defines partial ordering among tasks

T1 T3

T4 T2

22 EE-522: Embedded Systems 1/25/2018

 © Adeel Pasha, LUMS, Spring 2018

Important Notions
 Data Sharing
 Tasks often need to share data
 Example: A temperature sensing and processing node
 Data sharing can imply a task precedence
 But not mandatory
 Task precedence without data sharing or data sharing without
precedence?
T1 T3

T4 T2
23 EE-522: Embedded Systems 1/25/2018
 © Adeel Pasha, LUMS, Spring 2017

Tasks Classification

24 EE-522: Embedded Systems 1/25/2018