Event-based Deadline Miss Rate Simulator

Environment:

• Python 3.6

Contributions of this patch:

• conversion from Python 2.7 to 3.6
• extended uni-processor to multiprocessor environment
• implemented RP, EDF, RM and DM scheduling

Description of the adopted files from [1] and [2]:

• EPST.py contains the analyses of the upper bound of the deadline misses.
• bounds.py contains different bounds related to the Chernoff bound.
• task-generator.py contains the task generating routines (it is enhanced by [2] files).
• sort_task_set.py contains the task generating routines.
• TDA.py contains some time demand analyses (it is enhanced by [2] files).
• deadline_miss_probability.py contains the methods from [2].

The proposed methods and the simulator files:

• experiments.py contains the main function to run the simulator and the evaluations.
• multiprocessor_simulator.py / simulator.py contains the class of the event-based simulator.

How to use?

To run the simulator, run the following command from the repository:

python3 experiments.py [mode] [# tasks] [tasksets_amount] [generationType] [part]

mode:

0: generates tasksets using the other configuration parameters
2: starts a simulation with all combinations of the current configuration and saves the miss rate for each in the outputs folder
3: creates a plot for the current configuration
4: takes the current configuration and calculates the average amount of releases of all tasks for the generated taskset required, to reach the specified jobnumber
7: for testing purposes

# tasks:

desired number of tasks

tasksets_amount:

desired number of tasksets

generationType:

0: preset of possible periods (1, 2, 5, 10, 50, 100, 250 and 1000)
1: random periods between 1 and 100

part:

allows saving different tasksets with identical configuration

in-file configurations:

The scheduling method and processor type have to be set in experiments.py (lists 'schedulingMethods' and 'processorTypes').

processorType:

0: single
1: partitioned
2: global

scheduling:

1: random priority
2: earliest deadline first
3: deadline-monotonic
4: rate-monotonic

Reference

• [1] K. H. Chen and J. J. Chen, "Probabilistic schedulability tests for uniprocessor fixed-priority scheduling under soft errors", 2017 12th IEEE International Symposium on Industrial Embedded Systems (SIES), Toulouse, France, 2017, pp. 1-8.
• [2] Georg von der Brüggen, Nico Piatkowski, Kuan-Hsun Chen, Jian-Jia Chen, Katharina Morik, " Efficiently Approximating the Probability of Deadline Misses in Real-Time Systems", accepted in ECRTS 2018.