ERT-CORE MATLAB scripts (time complexity estimation and measurement)

Citation Author(s):
Ivan
Eroshkin
Czech Technical University in Prague
Submitted by:
Ivan Eroshkin
Last updated:
Tue, 05/17/2022 - 22:18
DOI:
10.21227/fj69-wv35
Data Format:
Link to Paper:
License:
0
0 ratings - Please login to submit your rating.

Abstract 

Resource Efficient Real-Time Reliability Model for Multi-Agent IoT Systems

Resource Efficient Real-Time Reliability Model for Multi-Agent IoT Systems is called ERT-CORE. It defines specific input parameters, i.e., system's workload, average request processing time and availability. Defined parameters reflect system's state and react on its changes. Based on these parameters system reliability evaluation is performed.

It is resource-efficient model and has a linear time complexity, which allows it to be used for real-time system reliability evaluation for monitoring purposes.

These MATLAB scripts perform time complexity estimation of three implemented algorithms and also perform an experiment/simulation with ERT-CORE model (show it's behavior).

Instructions: 

Time Complexity evaluation

Scripts for time complexity evaluation can be found under time_complexity/ folder. Please run main.m to replicate this experiment.

There are also snapshots of the Matlab environment, i.e. finalComputation1.mat, which could be reused in order to produce time complexity graphs. Time complexity evaluation on 3.7 GHz CPU could take up to one week.

In the time complexity estimation, a pseudo-random se-quence of numbers is generated to feed an input data for ERT-CORE and Reliability of a Thing [1] models. Only Reliability-Aware IoT model [2] requires pseudo-random Poisson number distribution as an input data. To exclude an operating system resource usage deviations from the measurement, a 50 000 iterations for each reliability model computations with different number of the systems in the range of〈0, 1000〉are performed and an average time is taken.

Measurement

Script for measurement performed in publication can be found under parameter_behavior/ folder. Please run sampleParameterBehaviorRandom_nv.m to replicate this experiment.

In the experiment, a set of user’s agents run in the system. Time periods when specific agent is operational in the simulation are following:

 

  • Agent 1 runs from 25th to 50th second,
  • Agent 2 from 70th to 100th second,
  • Agent 3 from 130th to 145th second,
  • and Agent 4 from 160th to 185th second.