SĆStefan Ćirković

Stefan Ćirković

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Stefan Ćirković has a total of 11 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Traffic Sign Recognition in Autonomous Vehicles Using Edge-Enabled Federated Learning

    by Aleksa Iričanin, Veljko Lončarević, Stefan Ćirković, Vladimir Mladenović

  • Protection of Web Servers from DDoS Attacks Originating from the Tor Network: Detection and Mitigation Using Tor's Hammer

    by Stefan Ćirković, Saša Cvetković

    Distributed Denial-of-Service (DDoS) attacks remain a serious threat to web servers, causing significant consequences. The Tor network, designed for privacy, is increasingly being exploited for launching DDoS attacks. This paper examines the challenges of detecting and mitigating DDoS attacks originating from the Tor network. It analyzes the characteristics of Tor that make it attractive to attackers, existing detection and mitigation methods, and the integration of protective mechanisms-rate limiting, iptables, and blocking Tor exit nodes-to reduce the impact of such attacks. The combination of these techniques for improved protection is considered, and their effectiveness in defending against Tor-based DDoS attacks is evaluated. It has been observed that many websites lack even basic defense mechanisms and that even fundamental DDoS protection can be beneficial.

  • A Computationally Implemented Method for Assessing Air Pollution in Urban Traffic Based on Traffic Density Analysis

    by Stefan Ćirković, Nikola Stanić, Katarina Karić

    Monitoring and assessing air pollution levels in urban areas represents a key challenge in environmental protection and traffic management. This paper proposes a computationally implemented method for indirect air pollution assessment based on traffic density analysis, eliminating the need for direct sensor-based measurements of pollutant concentrations. The proposed system employs computer vision for real-time vehicle detection and classification, while air pollution levels are estimated using a mathematical model that accounts for the average emission rates per vehicle. The system architecture includes an embedded computer (Raspberry Pi) for video data processing, network infrastructure connectivity, and data transmission to a central server for further analysis and visualization. Experimental results confirm that the proposed approach enables reliable and efficient air pollution assessment under real-world conditions, opening avenues for enhancing environmental monitoring strategies and optimizing traffic flow in urban environments.

  • Softversko rešenje za vizuelizaciju, grupisanje i filtriranje saobraćajnih nezgoda

    by Fakultet Tehničkih Nauka Univerzitet U Kragujevcu, Stefan Ćirković, Marija Blagojević, Milan Ilić, Beograd Agencija Za Bezbednost Saobraćaja, Srećko Ćurčić

    Cilj ovog rada je razvoj veb aplikacije koja omogućava preciznu identifikaciju kritičnih tačaka na saobraćajnoj mreži. Analizirani su ključni faktori koji utiču na nastanak saobraćajnih nezgoda. Kroz sprovedenu analizu trendova i obrazaca ponašanja učesnika u saobraćaju, istraženi su faktori kao što su sezonske, vremenske i prostorne varijacije učestalosti nezgoda. Dobijeni rezultati pružaju uvid u najkritičnije dane, mesece i uslove, kao i ulogu osvetljenosti na povećanje rizika tokom noći. Ovi rezultati mogu poslužiti kao osnova za kreiranje preventivnih strategija za obezbeđenje bezbednosti. Predloženi metodološki pristup omogućava primenu u različitim oblastima istraživanja, kao što su napredne tehnike mašinskog učenja i razvoj modela za predikciju nezgoda.

  • Analysis of Approaches to Developing Kotlin Multiplatform Applications and Their Impact on Software Engineering

    by Nikola Stanić, Stefan Ćirković

    This paper explores the concept of Kotlin Multiplatform and analyzes the approach to developing multiplatform applications using this tool. The research aims to analyze the key features of Kotlin Multiplatform, including its ability to share code across different platforms such as Android, iOS, and web. Through a detailed analysis of existing literature and case studies, the paper will explore the advantages and challenges of developing multiplatform applications using Kotlin Multiplatform, as well as their impact on software engineering. Special attention will be paid to performance, scalability, and code management within Kotlin Multiplatform projects. Based on the gathered data, the paper will also explore future perspectives of multiplatform application development, including potential trends, technological advancements and other cross platform solutions.


  • Application of the YOLO algorithm for Medical Purposes in the Detection of Skin Cancer

    by Stefan Ćirković, Nikola Stanić

    Skin cancer is one of the most common forms of cancer worldwide. Exposure to ultraviolet (UV) radiation increases the risk of its development. Early preventive examinations and early detection of suspicious skin changes are key factors for successful treatment. Due to the rapid development of AI technologies, neural networks have found application in various fields, including medicine. Neural networks can be used to create various applications, which would facilitate self-examination for patients and alert them to potential problems. This method would further save time and reduce healthcare costs. The paper presents the application of a neural network using the YOLO (You Only Look Once) algorithm on a dataset of mole images with the aim of identifying and classifying moles, which facilitates early intervention and improves treatment outcomes.