Efficient Analysis to Provide Weakly-Hard Real-Time Guarantees · Zain A. H. Hammadeh . Efficient...

Efficient Analysis to Provide Weakly-Hard Real-Time Guarantees

Zain A. H. Hammadeh

Efficient Analysis to Provide Weakly-Hard Real-Time Guarantees

• Real-Time System?

• Hard Real-Time System?

• Weakly-Hard Real-Time System?

• Weakly-Hard Real-Time Guarantees?

• Efficient Analysis?

2

Primitive Definitions

3

In real-time systems the correctness of the system depends not only on the functional results of the computations, but also on the time at which the results are produced

sensor

interrupt

engine


4


sensor

interrupt

engine

Void task1 () {

#task_body; }

Void task2 () {

#task_body; }

Void task3 () {

#task_body; }


5


sensor

interrupt

engine

Void task1 () {

#task_body; }

Void task2 () {

#task_body; }

Void task3 () {

#task_body; }

Task is the executable unite of work scheduled by the operating system


6


sensor

interrupt

engine

Void task1 () {

#task_body; }

Void task2 () {

#task_body; }

Void task3 () {

#task_body; }


7


sensor

interrupt

engine

Scheduler


8


sensor

interrupt

engine

Scheduler

Scheduling policy determines how the contention between tasks is resolved


9

Time

Interrupt

Sensor

Engine


10

Time

Interrupt

Sensor

Engine


11

Time

Activation Interrupt

Sensor

Engine


12

Time


Sensor

Engine

Period


13

Time


Sensor

Engine

Period


14

Time

Activation

Termination

Interrupt

Sensor

Engine

Period


15

Time

Activation

Termination Execution time

Interrupt

Sensor

Engine

Period


16

Time

Activation


Interrupt

Sensor

Engine

Period


17

Time

Activation


Priority

Interrupt

Sensor

Engine

Period


18

Time

Activation


Priority

Interrupt

Sensor

Engine

Period


19

Time

Activation


Priority

Interrupt

Sensor

Engine

Period


20

Time

Activation


Priority

Interrupt

Sensor

Engine

Period


21

Time

Activation

Termination

Preemption

Execution time

Priority

Interrupt

Sensor

Engine

Period


22

Time

Activation

Termination

Preemption

Execution time

Priority

Interrupt

Sensor

Engine

Period


23

Time

Activation

Termination

Preemption Resume

Execution time

Priority

Interrupt

Sensor

Engine

Period


24

Time

Activation

Termination

Preemption Resume

Execution time

Priority

Interrupt

Sensor

Engine

Period


25

Time

Activation

Termination

Preemption Resume

Execution time

Priority

Interrupt

Sensor

Engine

Period


26

Time

Activation

Termination

Preemption Resume

Execution time

Priority

Interrupt

Sensor

Engine

Static Priority Preemptive SPP

Period

Scheduling Policy

• Static priority preemptive SPP

• Static priority non-preemptive SPNP

• Earliest deadline first EDF

• Time division multiple access TDMA

• Round-Robin RR

27


28

Time

Activation

Termination

Preemption Resume

Execution time

Priority

Interrupt

Sensor

Engine


Period


29

Time

Activation

Termination

Preemption Resume

Execution time Response time

Priority

Interrupt

Sensor

Engine


Period


30

Time

Activation

Termination

Preemption Resume

Execution time

Deadline

Response time

Priority

Interrupt

Sensor

Engine


Period


31

Time

Activation

Termination

Preemption Resume

Execution time

Deadline

Response time

Priority

Interrupt

Sensor

Engine


Period


32

Time

Activation

Termination

Preemption Resume

Execution time

Deadline

Response time

Priority

Interrupt

Sensor

Engine


Period

Hard vs. Weakly-Hard

• Response time <= Deadline • Hard real-time systems

– All deadlines have to be met – Safety-critical systems like ABS are Hard real-time

systems

• Weakly-hard real-time systems – The distribution of deadline misses/met are precisely

bounded – Control system, Image processing are Weakly-hard

real-time systems

• Others: Firm and Soft

33

Efficient analysis for real-times guarantees

Hard real-time

• It should cover performance corner-cases – Worst-case scenario

• Worst-case response time

Weakly-hard real-time

• It should cover performance corner-cases – Worst-case scenario

• Worst-case response time

• Bounds on the distribution of deadline misses

34

Worst-Case Response Time Analysis

35

1 1 1

1

3

2

2

𝝉𝟏

𝝉𝟐

𝝉𝟑

𝝉𝟒 1

1

3

Time(ms) 0 10

SPP


36

1 1 1

1

3

2

2

𝝉𝟏

𝝉𝟐

𝝉𝟑

𝝉𝟒 1

1

3

Time(ms) 0 10

WCRT4 = 16ms

SPP


37

1 1 1

1

3

2

2

𝝉𝟏

𝝉𝟐

𝝉𝟑

𝝉𝟒 1

1

3

Time(ms) 0 10 largest level-4 busy window

WCRT4 = 16ms

SPP


38

1 1 1

1

3

2

2

𝝉𝟏

𝝉𝟐

𝝉𝟑

𝝉𝟒 1

1

3


WCRT4 = 16ms

SPP

𝑫𝟒 = 𝟏𝟐ms


39

1 1 1

1

3

2

2

𝝉𝟏

𝝉𝟐

𝝉𝟑

𝝉𝟒 1

1

3


WCRT4 = 16ms

SPP

𝑫𝟒 = 𝟏𝟐ms

Weakly-Hard Guarantees

40

It is described in a form of m-out-of-k (m,k) Example: (2,5), (1,2)

Weakly-Hard Guarantees

41

It is described in a form of m-out-of-k (m,k) Example: (2,5), (1,2)

The distribution of deadline misses/met?

The Deadline Miss Model for a task 𝜏𝑖 is a function

𝑑𝑚𝑚𝑖: ℕ+ → ℕ+ such that 𝑑𝑚𝑚𝑖(𝑘)

• bounds the maximum number of executions of 𝜏𝑖 which may have a response time larger than 𝐷𝑖 in a window of k consecutive executions of 𝜏𝑖

Deadline Miss Model

42

In many practical cases, embedded system applications accept occasional deadline violations as long as their number can be bounded

• I focus on bounding the distribution of deadline misses in a form of (m,k)-guarantees for independent weakly-hard real-time tasks and task chains using TWCA.

• Exploring the design space for (m,k)-schedulability is a part of my research as well.

Summary

43

Publications

• Zain A. H. Hammadeh, Sophie Quinton, and Rolf Ernst, "Extending Typical Worst-Case Analysis Using Response-Time Dependencies to Bound Deadline Misses" in Proceedings of the 14th International Conference on Embedded Software, (New Delhi, India), October 2014.

• Wenbo Xu, Zain A. H. Hammadeh, Sophie Quinton, Alexander Kr öller, and Rolf Ernst, "Improved Deadline Miss Models for Real-Time Systems using Typical Worst-Case Analysis" in 27th Euromicro Conference on Real-Time Systems, (Lund, Sweden), July 2015.

• Leonie Ahrendts, Zain A. H. Hammadeh, and Rolf Ernst, "Guarantees for Runnable Entities with Heterogeneous Real-Time Requirements" in Design, Automation & Test in Europe Conference & Exhibition (DATE), March 2016.

• Zain A. H. Hammadeh, Sophie Quinton, Rafik Henia, Laurent Rioux, and Rolf Ernst, "Bounding Deadline Misses in Weakly-Hard Real-Time Systems with Task Dependencies" in Design Automation and Test in Europe (DATE), (Lausanne, Switzerland), March 2017.

44

Publications • Rafik Henia, Laurent Rioux, Nicolas Sordon, Zain A. H. Hammadeh, Sophie Quinton,

and Rolf Ernst, "Demo Abstract: Bounding Deadline Misses for Weakly-Hard Real-Time Systems Designed in CAPELLA" in The 2017 IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS) - Demo Track, (Pittsburgh, PA, USA), April 2017

• Zain A. H. Hammadeh, Sophie Quinton, Marco Panunzio, Rafik Henia, Laurent Rioux, and Rolf Ernst, "Budgeting Under-specified Tasks for Weakly-Hard Real-Time Systems" in The 29th Euromicro Conference on Real-Time Systems (ECRTS17), (Dubrovnik, Croatia), June 2017.

• Mohammad Hamad, Zain A. H. Hammadeh, Selma Saidi, Vassilis Prevelakis und Rolf Ernst, "Prediction of Abnormal Temporal Behavior in Real-Time Systems" in The 33rd ACM/SIGAPP Symposium On Applied Computing (SAC 2018), (Pau, France), April 2018.

45

Efficient Analysis to Provide Weakly-Hard Real-Time Guarantees · Zain A. H. Hammadeh . Efficient...

Documents

Transcript of Efficient Analysis to Provide Weakly-Hard Real-Time Guarantees · Zain A. H. Hammadeh . Efficient...