Jury work

THE REPORT

About deciding on the “Blažo Mirčevski” Award

During the submission of works to TELFOR 2022, 20 works by young authors (up to 30 years of age) were registered that had an average reviewer’s rating greater than or equal to 3. Of these, 4 works carried the reviewers’ proposal for awarding the “Blažo Mirčevski” Award. 4 works were exhibited online, DP – 2.8., DP – 4.14., DP – 5.19. and DP – 7.5. and the other 16 were presented by the authors immediately present.

Young authors were most interested in the field of Signal processing, where there were 6 papers, in the Software Tools and Applications section there were 4 papers, in the Communication Systems and Applied Electronics sections there were 3 papers each, in the Telecommunication Networks section – 2 of work and one each in the Applied Electromagnetics and Multimedia sections.

All 20 works of young authors in the competition for the “Blažo Mirčevski” Award were written in English, which also applied to presentations. The degree of success of the presentation was very different. Some authors had presentations in fluent English but they stuck too much to the presentation, sometimes they just read. The level of understanding of the questions and answers was also a problem for many presenters. The quality of the prepared presentations was also very different, from a simple copy of the text of the paper to presentations with color slides that seem informative and documented and form the basis for a better understanding of the paper. The works exhibited in the online mode had a considerable level of presentation and left the impression that the young authors are very competent. Of the 4 works exhibited online, two were nominated for the “Blažo Mirčevski” Award.

Some interesting areas of research that were included in the works are features of the general trend in modern technologies in recent times. There were works dealing with smart city, deep learning, secure multiparty computation, textile antennas, radar systems. However, the depth of understanding the problem and the effectiveness of the solutions offered by the authors are different.

The number of works selected for quality was gradually narrowed down, and finally the jury agreed that the following three works will be shortlisted for the “Blažo Mirčevski” Award for TELFOR 2022:

1. Paper 2.8. Capacity Analysis of Channel Models for Different Environments at 28/39 GHz with NYUSIM by Ahmet Kola, Çetin Kurnaz

2. Paper 5.19. DOA Estimation in MIMO Radars via Deep Learning by Kerem Maden, Işın Erer

3. Paper 9.3. Smart city use case development based on FIWARE technology by Amera Sinanovic, Emir Meskovic, Aljo Mujcic, Nermin Suljanovic

Paper 2.8. Capacity Analysis of Channel Models for Different Environments at 28/39 GHz with NYUSIM, originates from the Department of Electrical and Electronics Engineering, Samsun University

The paper investigates what influences the communication efficiency of the 5G network depending on the communication environment. Channel capacity was investigated using the NIUSIM channel simulator. NIUSIM is a GUI file created using MATLAB software. The work resulted in an analysis of which channel parameters can be used to optimize communication efficiency in different environments

In essence, the paper deals only with simulation without any attempt at realization.

The references given are even from 2021, so you can see a good understanding of the field.

Paper 5.19. DOA Estimation in MIMO Radars via Deep Learning by Kerem Maden, Işın Erer

The paper proposes a new architecture – DCAE-CNN architecture, which consists of convolutional autoencoders (Denoising Convolutional AutoEncoders – DCAE) and convolutional neural networks (CNN).

The proposed architecture is well explained and illustrated with a detailed diagram that clearly shows the places of execution of certain functions and the steps in signal processing. The experimental results are explained and a comparison with the existing standard algorithm is made. Simulation results show that optimized networks perform better than conventional algorithms in the task of DOA estimation. The algorithm proposed in the paper is approximately 6 times faster than the traditional MUZIC algorithm in conditions of low SNR (Signal Noise Rating).

As part of the treatment of experimental results, an overview of the hardware and software platform on which the DCAE-CNN architecture was built (Python using Tensorflow), neural network training and testing was performed. Computer parameters (processor, memory, clock pulse) are explicitly stated. These data are very important for evaluating the success of the proposed algorithm and are rarely mentioned in similar works, so it is often not possible to evaluate the impact on the calculation time – algorithm or computer performance.

The conclusion is good but very simple without considering any future work. Claims from the abstract are repeated a little. The authors note that training a neural network on a significant amount of data and optimizing hyperparameters for specific antenna arrays requires a certain amount of time.

The references are very diverse, from the well-known textbook for radar systems, MIMO Radar Signal Processing to the latest papers in IEEE Communications Letters. Since detailed theoretical explanations are given in the paper with the use of extensive mathematical apparatus, it is evident that he has a good knowledge of the literature in the subject area.

Paper 9.3. Smart city use case development based on FIWARE technology

The paper originates from the Faculty of Electrical Engineering, University of Tuzla, Bosnia and Herzegovina

The paper deals with the very current topic of “smart city” by presenting the FIWARE smart city model that could be implemented in environments that already exist and possess some smart city components. A list of technical specifications that the city administration needs to fulfill in order to be able to implement a smart city platform based on FIWARE is given.

The authors considered a use case for the FIWARE platform using the platform’s out-of-the-box capabilities. The paper does not show that anything specific has been developed (eg some plug-in), so it can be concluded that this tool has not actually been applied in any real situation. One gets the impression that the entire work is more at the level of an overview of the platform’s capabilities.

This work is characterized by a very good presentation by the author Amera Sinanović, which indicates the author’s complete involvement in the realization of the work.

After all consideration and consultation, the jury unanimously decided to declare the work of 5.19 as the best work of a young author at the TELFOR 2022 conference. DOA Estimation in MIMO Radars via Deep Learning and the “Blažo Mirčevski” Award was won by the young author Kerem Maden.

The author’s contact information is as follows: Kerem Maden, Radar, Electronic Warfare and Intelligence Systems Division ASELSAN, Ankara, Turkiye, kmaden@aselsan.com.tr

The jury was composed of:

Vladimir Orlić, PhD
Prof. dr Marija Malnar, PhD
Nemanja Zdravković, PhD
Nikola Popović, M.Sc.
Mihailo Tošović, B.Sc.
Julijana Mirčevski, B.Sc.

During the various phases of the evaluation for individual works, the following consultations were used:

Ljiljana Marijanović, PhD, A1 Digital, Austria
Mitar Simić, PhD, FTN Novi Sad
Slavica Tomović, PhD, Elektrotehnički fakultet, Podgorica
Vuk Batanović, PhD, ETF Beograd

Julijana Mirčevski

Belgrade, May 17, 2022