The aim of IEEE Time-Sensitive Networking (TSN) standards is to grant deterministic communication in traditional Ethernet networks for Industry 4.0. Insofar as the use cases in the Factory need some mobility, the extension of the TSN capabilities over the fifth-generation (5G) cellular network is the next step. Some challenges in TSN over 5G, such as TSN translators time synchronization functionality, are well defined in the standards, even if they have not yet been addressed in the market. However other challenges, such as the dynamic configuration of the entire network (or part of the it) based on quality requirements of the current TSN traffic pattern, are defined at a very high level and delegated to vendors for implementation. This paper addresses this challenge, using an Automata Learning approach to monitor and reconfigure the end-to-end 5G QoS flow to keep the quality of a TSN session within the required values. Additionally, algorithms are provided to build the automata from network data and predict potential deviations of the requirements to meet the expected quality. Moreover, this work presents a functional TSN over a 5G testbed where the algorithms have been tested, demonstrating that the proposed solution achieves an improvement of around 40% compared to the usual operation of the network.

In this paper, we study the adequacy of applying peer-to- peer techniques to mobile networks by conducting a series of experiments using smart phones as peers. We measure important parameters, such as jitter and packet losses, in static and dynamic scenarios, focusing on a video streaming service. Finally, based on the results obtained, we discuss the feasibility of these applications.

The evolution of mobile networks and the increasing number of scenarios for mobile applications requires new approaches to ensure their quality and performance. The TRIANGLE project aims to develop an integrated testing framework that allows the evaluation of applications and devices in different network scenarios. This paper focuses on the generation of user interactions that will be part of the test cases for applications. We propose a method that combines model-based testing and guided search, based on the Key Performance Indicators to be measured, and we have evaluated our proposal with an example. Our ultimate goal is to integrate the guided generation of user flows into the TRIANGLE testing framework to automatically generate and execute test cases.