Sava Stanišić

Sava Stanišić

Auto-generated profile, based on publicly available information

Sava Stanišić

University of Kragujevac, Cacak, Serbia

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 June 2025
Verified
Conference Paper
Closed access
SSSava Stanišić
Sava Stanišić
NZNikola Zogović
Nikola Zogović
Selecting cost-efficient and highly available cloud infrastructure is a critical challenge for modern applications. This paper proposes a multi-objective optimization framework leveraging genetic algorithms (Non-dominated Sorting Genetic Algorithm II - NSGA-II) to model the trade-off between two conflicting objectives, infrastructure costs and service availability. By analyzing the Pareto front of non-dominated solutions, the optimal virtual machine (VM) configurations are identified under a budget constraint. Experimental results demonstrate that our approach outperforms single-objective strategies in balancing cost and availability, achieving up to $\mathbf{9 9. 5 \%}$ service availability at a cost of $\mathbf{$ 0. 5 5 / h}$.

1 January 2024 in Vojnotehnicki glasnik
Verified
Journal Article
Open access
PDF file
SSSava S. Stanišić
Sava S. Stanišić
HSHristina N. Stojanović
Hristina N. Stojanović
IĐIgor Lj Đorđević
Igor Lj Đorđević
Introduction/purpose: This work provides a comprehensive overview of blockchain technology, elucidating its foundational principles and how it ensures transparency, immutability, and decentralization. The integration of Solidity with blockchain is explored through theoretical approach. Methods: This work meticulously dissects blockchain principles, elucidating transparency, immutability, and decentralization, while exploring Solidity integration in a theoretical framework, ensuring a comprehensive understanding of their intricate relationship and contributing to a broader comprehension of modern distributed ledger technology. Results: The resulting product of this paper will be getting useful knowledge about the technology that practically shapes the world. Conclusion: In conclusion, the adoption of Solidity as a programming language in blockchain technology has proven to be pivotal, enhancing smart contract functionality and overall system security. Its specialized features make it an indispensable tool for developers navigating the complexities of decentralized applications.

1 January 2024 in Vojnotehnicki glasnik
Verified
Journal Article
Open access
PDF file
MMMarko R. Marković
Marko R. Marković
SIStefan M. Ivanović
Stefan M. Ivanović
SSSava S. Stanišić
Sava S. Stanišić
Introduction/purpose: This paper conducts a comprehensive analysis of a potential implementation of Voice over Internet Protocol (VoIP) systems, focusing on network architecture, VoIP phones, and servers. The study explores potential vulnerabilities and proposes solutions. The paper concludes by advocating for a holistic approach to securing VoIP systems, incorporating supplementary services to ensure the confidentiality, integrity, and availability of voice communications in the digital landscape. Methods: Review of the underlying theory, analysis of the end-user needs and potential solutions, practical viability assesment. Results: The theoretical points discussed were proven in practice, using commercially available resources. Communication was established in an expected manner. Conclusions: Implementing solutions similar to the one presented in the paper would be a relatively inexpensive way to make diverse improvements to the operation of tactical level units, both in peacetime and during war.

1 January 2025
Verified
Conference Paper
Closed access
SSSava Stanisic
Sava Stanisic
VMVladimir Mladenovic
Vladimir Mladenovic
Managing network traffic and ensuring security has become critical for maintaining robust Kubernetes environments. This paper aims to investigate the implementation of Kubernetes Network Policies to enhance network security and operational efficacy. The objectives of these research include exploring the configuration of NetworkPolicy resources, employing label selectors, and managing namespaces. The methodology presented in this paper involves leveraging advanced tools like Flannel and Calico to enforce network policies effectively. The results indicate significant improvements in traffic control and workload security, offering a comprehensive guide for optimizing Kubernetes clusters with refined network policy management techniques.

1 January 2025 in Tehnika
Verified
Journal Article
Open access
PDF file
SSSava Stanišić
Sava Stanišić
BĐBorislav Đorđević
Borislav Đorđević
ITIvan Tot
Ivan Tot
OROlga Ristić
Olga Ristić
In modern container technologies, Docker is a key tool for deploying and managing applications. Understanding the performance of different Docker networking modes is crucial for efficient system design. This paper analyzes the performance of three core Docker network modes-Bridge, Overlay, and MacvLAN-in a production environment (Docker 27.5.0, Ubuntu 24.04.2 LTS). The methodology involves measuring key performance indicators-latency and throughput-under varying workloads. Results show that MacvLAN, with direct access to the physical interface, delivers the best performance (0.3 ms latency, 9.8 Gbps throughput) but requires strict IP allocation control. Overlay networks, using VXLAN encapsulation, introduce ~1.1 ms additional latency per hop and require careful MTU tuning (≤1400 bytes) to prevent packet fragmentation. Bridge mode shows limitations under high-throughput scenarios due to DNAT processing, leading to ~14% packet loss with a standard MTU of 1500. In conclusion, the optimal Docker network mode depends on workload characteristics: MacvLAN is ideal for latency-sensitive applications, Overlay suits distributed systems with optimized MTU settings, and Bridge is best for development environments when paired with an external load balancer.

28 April 2026 in JUCS - Journal of Universal Computer Science
Verified
Journal Article
Open access
PDF file
SSSava Stanisic
Sava Stanisic
BĐBorislav Đorđević
Borislav Đorđević
BBBranislav Belotic
Branislav Belotic
OROlga Ristić
Olga Ristić
ITIvan Tot
Ivan Tot
DKDimitrije Kolasinac
Dimitrije Kolasinac
He orchestration of microservices in distributed cloud environments poses significant challenges due to the heterogeneous nature of cluster nodes and dynamic workload patterns. Traditional scheduling strategies in Kubernetes often fail to optimize latency-sensitive applications effectively. This paper proposes a latency-aware orchestration framework that integrates reinforcement learning techniques to dynamically schedule and migrate microservices across heterogeneous Kubernetes clusters. The proposed approach leverages a deep Q-network (DQN) agent trained to minimize end-to-end response times while balancing resource utilization and avoiding service-level objective (SLO) violations. Experiments conducted on a hybrid testbed comprising virtual and physical nodes demonstrate that the reinforcement learning-based scheduler reduces latency by up to 25% compared to default Kubernetes scheduling policies. The results highlight the potential of intelligent orchestration methods to enhance performance in complex cloud-native deployments.