Ing. Jaroslav Frnda, Ph.D., works at the Faculty of Operation and Economics of Transport and Communications of the University of Žilina. He focuses on the quality of multimedia services transmitted through next-generation access networks, machine learning algorithms, and data analysis.
At the turn of the new millennium, transmission networks became digitalised and converged. Nowadays, people use multiple mobile devices, and they can call, scroll through social networks or watch TV at the same time. This also means that television, the internet, and the fixed (telephone) line no longer operate on separate transmission channels but are transmitted together. With increased congestion on the transmission capacity, it is necessary to evaluate the quality of image and voice services continuously.
Jaroslav Frnda deals with methods and metrics that evaluate the quality of multimedia services transmission. Today, he focuses most on video quality – mainly due to the popularity of streaming services. Playing online videos depends on the quality of the Internet transmission and IP protocol, and there may be multiple errors during transmission. Delays in the network or lost data can impact the image quality perceived by the end customer. Video quality may deteriorate, experiencing blips or becoming blurred or pixelated.
"Evaluation is done for a variety of reasons. The first is for online streaming service providers to know whether they are providing a competitive service. Secondly, so that customers know when they have the right to complain, and the regulatory authority can check such complaints based on an objective metric. Finally, research teams working on developing new encoding mechanisms for video need to be able to evaluate their designs against the currently available solutions objectively," explains Jaroslav Frnda. Usually, a similar evaluation is made on the subjective assessment of the research group. "According to the standardized metric of the International Telecommunication Union, there must be at least fifteen people watching a video preview to evaluate its quality from 1 to 5. If the transmission parameters in the network deteriorate for the operator, it is difficult to call so many people for evaluation, and it does not work in real-time."
Jaroslav Frnda and his team work on an objective method based on mathematical regulations, which simulates this process, also called the "human visual system." In real-time, it will evaluate the damaged section of the video and, based on the machine learning and data algorithm, assess how human subjects would react to a similar demonstration. At the same time, subjective testing results in the repeated measurement of deviations, Jaroslav Frnda's methodology produces consistent video quality results
A native of Martin, he developed a relationship with technology during his physics class in high school. He studied informatics and communication technologies at the Technical University of Ostrava, in the Czech Republic. Today he works at the Faculty of Operations and Economics of Transport and Communications of the University of Žilina. He claims that the most exciting part of his job is the freedom to research and connect with scientists worldwide. He recently published an article with international colleagues in the journal Scientific Reports, which is part of the prestigious publishing house that publishes Nature. Jaroslav Frnda also considers receiving a grant from Slovakia’s Scientific Grant Agency (VEGA) in 2021 as one of his achievements.
Jaroslav Frnda is an active person and says that his hobbies balance his rigid work in science – he likes to cycle, go swimming, and used to play hockey. He is also interested in public affairs.