Prof. Dr. Ivan Kaštelan

Ivan Kaštelan received the BSc, MSc and PhD degrees in electrical and computer engineering from the Faculty of Technical Sciences, University of Novi Sad, Serbia in 2008, 2009 and 2014 respectively. He received the award for the best student of the Faculty of Technical Sciences in 2008 and the Faculty of Technical Sciences award “Momčilo Momo Novković“ for his teaching work in 2014. He is currently an Associate Professor at the Department of Computing and System Control at the University of Novi Sad, Faculty of Technical Sciences, teaching courses in digital and computer system design and the design of algorithms. As an instructor at NIT Institute, he is teaching the Python programming courses. He participated in multiple European (FP7, Horizon 2020, IPA) and national projects. He was the deputy coordinator of the E2LP project that developed the learning platform used in 8 higher education institutions across the European continent. Currently he is a coordinator of the project funded by the Government of AP Vojvodina in the field of applicability of renewable energy sources. He published 7 papers in international journals and more than 80 papers in the conference proceedings. He is also a co-author of 13 patents.

References

Željko Lukač, Ivan Kaštelan, Mario Vranješ, Branislav Todorović: "AMV ALPHA Learning Platform for Automotive Embedded Software Engineering", IEEE Transactions on Learning Technologies, 2021, Vol. 14 #3, pp. 292-298, https://doi.org/10.1109/TLT.2021.3098505
Ivan Kaštelan, Vukota Peković, Nikola Teslić: "A Novel Concept of Electrical Stimulation of Touchscreens Used for Automated Verification of Mobile Devices", Elektronika ir Elektrotechnika, 2019, Vol. 25 #2, pp. 47-52, http://dx.doi.org/10.5755/j01.eie.25.2.23204
Ivan Kaštelan, Jorge R. Lopez Benito, Enara Artetxe Gonzalez, Jan Piwinski, Moshe Barak, Miodrag Temerinac: "E2LP: A Unified Embedded Engineering Learning Platform", Microprocessors and Microsystems (Elsevier), 2014, Vol. 38 #8, pp. 933-946, http://dx.doi.org/10.1016/j.micpro.2014.09.003
Roman Szewczyk, Ivan Kaštelan, Miodrag Temerinac, Moshe Barak, Vlado Sruk: "Embedded Engineering Education", Springer, 2016, http://dx.doi.org/10.1007/978-3-319-27540-6
Milan Bjelica, Ivan Kaštelan, Maja Barić: "Sistem i metod za udaljenu nastavu kombinovanjem televizije i interneta" (English: "System and Method for Remote Teaching Combining the TV and Internet"), Patent application RS P-2020/1039

Courses

Advanced Python programming

200

Python programming

200

Fault Tolerant Systems

249

Learn how to make your systems work under unexpected and faulty conditions.

With this course, you will be introduced to the basics of fault tolerance. You will learn about the concepts of design and implementation of fault tolerance mechanisms in general systems are introduced. You’ll get to know the quantitative and qualitative methodology which is used in the evaluation of specific fault tolerance principles. After the course, you will be able to understand and perform key alterations to system functions, components, or mechanisms. You will also be able to add an appropriate level of redundancy with the purpose of achieving the expected reliability of systems, despite the residual faults in the system design.

Course topics:

Main system dependability concepts, correlation with system safety and functional safety;

System integrity and safety integrity concepts;

Threats to system dependability (such as faults, errors, or failures);

Basic analysis of high-level composite system diagrams, key areas for dependability improvement;

Fault tolerance mechanism in appropriate stages (error detection, damage assessment, error recovery, fault treatment);

Static, dynamic, and hybrid redundancy calculations;

Fault-tolerant information exchange;

Fault tolerance mechanisms, such as channel and error models, detection/correction codes, and retry mechanisms;

Fault tolerance and system repair.

Modules:

M1: Dependable systems and fault tolerance

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M1.jpg

Provide the introduction to system dependability and the relevant building aspects of dependable systems (availability, reliability, safety, maintainability, integrity). The lecture gives an overview into the main concepts of fault tolerance, being that physical, informational or temporal fault tolerance, as well as the software fault tolerance as a specific consideration.

For the semi-autonomous vehicle, participants will discuss each term’s meaning, provide examples of failures, and identify techniques for fault prevention, removal, forecasting, and various fault tolerance approaches.

2 hours 30 minutes

M2: Static redundancy

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M2.jpg

Provide the introduction to fault tolerance techniques regarding the physical system architectures and the addition of redundant components in a static (always working) configuration. The lecture recaps the important random failures metrics (failure probability, reliability, reliability modeling via RBDs, failure rates and e-systems calculation), and gives insights to most common static redundancy configuration and the calculus regarding the improvement of reliability of systems modeled in such a way.

For the given system determine the system’s reliability and recalculate reliability with additional static redundancy for components.

2 hours 30 minutes

M3: Dynamic redundancy

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M3.jpg

The lecture discusses the potential of reconfiguring the system in the presence of faults/errors, during the system runtime. Specific mathematical modeling for such configurations is laid out, including stochastic processes (Markov chains) and how they can be used to model any dynamic behavior of the system experiencing failures, such as by activating spare components or performing the system repair.

For the given system adjust the reliability considering the individual component failure rates.

2 hours 30 minutes

M4: Standby systems

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M4.jpg

Standby systems are specifically focusing on dynamic redundancy application with hot and cold spare components, and their specific implementation practices. The lecture gives the required calculus for modeling standby systems and discusses gains in terms of reliability.

Participants will analyze how fault tolerance improvements were applied in the selected system from the industry of choice, including the feasibility of implementation of static redundancy mechanisms and evaluating the effects to reliability.

2 hours 30 minutes

M5: Repairable systems

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M5.jpg

Specific repair process and key repair terminology and variables are introduced (maintenance probability, repair rate). Systems are modeled for repair, and it is shown how the repair process may achieve the required system availability. Concepts of availability and repair are closely correlated, providing means to calculate the availability and reliability of the system and establish their relation in the fault tolerance context.

Participants will evaluate the allowed average repair time and complete the calculations according to the given parameters.

2 hours 30 minutes

M6: Physical fault tolerance applications (Project 1)

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M6.jpg

Physical fault tolerance concepts (static and dynamic redundancy and repair) are now applied to a specific example of a system from the relevant industry (e.g. automotive).

Participants are required to apply various fault-tolerance techniques to increase the dependability of the vehicle domain controller. The project is worked out in groups.

10 hours

M7: Information fault tolerance and block codes

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M7.jpg

Introducing the basic information fault tolerance concepts which are used to tolerate errors that happen during the information exchange, e.g., in the communication channel. Channel coding concepts are introduced, together with binary block codes, which can be constructed to add redundant data to the data being sent to later detect or correct the errors resulting from the fault in the communication channel.

Participants will model the reliability of typical communication channels in electronic systems, and see how errors affect the overall system failure rate, then reassess by applying basic error-correction codes.

2 hours 30 minutes

M8: Linear block codes

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M8.jpg

Deepening the information fault tolerance concepts with a specific class of codes allowing interesting properties which facilitate encoding and decoding process, including generator and parity check matrices, as well as syndrome decoding.

Each group will construct either an encoder or decoder with error detection for a system, using Hamming codes to tolerate bit errors, and then test and verify their solution with provided data.

2 hours 30 minutes

M9: Polynomial codes

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M9.jpg

A specific class of codes are introduced including all the relevant mathematical foundations required for understanding this class of codes and their application for channel encoding and decoding.

Each group will construct either an encoder or decoder with error detection for a system, using polynomial codes to tolerate bit errors, and then test and verify their solution with provided data.

2 hours 30 minutes

M10: Cyclic codes and CRC

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M10.jpg

A specific class of polynomial codes, bring in properties that allow the straightforward construction of the code and easy decoding and encoding implementation in simple hardware accelerators. A specific class of systematic cyclic codes allows the calculation of CRC checksums which are widely applied to detect errors that happen due to communication, unreliable storage media, etc.

Participants will design a fault-tolerant system using a cyclic code, including generator polynomial selection, encoding/decoding processes, and LFSR simulation.

2 hours 30 minutes

M11: Information fault tolerance applications (Project 2)

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M11.jpg

Brings in a real industry example where the information fault tolerance is applied together with the physical fault tolerance concepts. Students are required to follow their own system example in an assignment, to estimate the required channel code, construct it and demonstrate the dependability gains for their system.

Participants will need to implement information fault tolerance concepts on a vehicle domain controller to improve overall system dependability.

10 hours

M12: Temporal fault tolerance

https://academy.nit-institute.com/wp-content/uploads/2024/08/FTS-M12.jpg

Lecture addresses the application of time-based methods, such as repeat and compare, automatic repeat request (ARQ), sliding window and alike, which are based on repeating the operation to combat transient faults which lead to erroneous operation outputs. The lecture also gives the recap of all fault tolerance methods given in the course, by stating out their summary applicability for all important fault tolerance phases (error detection and correction, damage assessment and confinement, error recovery and fault treatment).

none

1 hour

Final exam

none

1 hour 30 minutes

Requirements

Software: Chrome browser.
Hardware: Computer with an Internet connection, working speakers, and microphone.
Prior knowledge: Students should have previous knowledge of selected calculus topics, such as basic differential equations, operations with matrices, Galois fields, vector spaces, and Boolean algebra. Students shall have basic knowledge of system and safety engineering and system reliability concepts, such as failure probability, reliability, failure rate, constant failure rate, MTTF, FIT, etc.