OPE2 – Operational & Experimental Insights #2
Wednesday, 17 June 2020, 12:15-14:30 CEST, Recommended re-viewing, https://www.youtube.com/playlist?list=PLjQu6nB1DfNCA18Qkp9876XJ_u6385TZi
Wednesday, 17 June 2020, 12:15-16:00 CEST, Non-Live interaction (Chat), link sent only to Registered people
Alex Mavromatis and Dimitra Simeonidou (University of Bristol, United Kingdom (Great Britain))
Smart applications enabled through Internet of Things (IoT), Artificial Intelligence (AI) and other fast growing technologies are forming a new market for all industries. As this new era of smart technologies is exciting, it can also can be daunting considering the huge amount of new software, hardware and networking concepts. A solution to adjust smoothly to the new era is experimenting over a testbed in order to learn, test and assess the feasibility of new applications. This paper is documenting the experiences and results of building an IoT testbed with an edge computing orientation. The testbed is designed to be adaptable to heterogeneity of devices, and deploy micro-services that are running “”close”” to the users. We present our software architecture for service orchestration, IoT data management and edge node management. The challenges of constructing a testbed outdoors in the city center of Bristol United Kingdom and our approach to solutions are thoroughly reported in this paper, aiming to contribute our lessons learned.
Min Xie (Telenor Research & Telenor Group, Norway); Joan Pujol-Roig (Samsung Electronics, United Kingdom (Great Britain)); Foivos Michelinakis (Simula Metropolitan, Norway); Thomas Dreibholz (Simula Metropolitan Centre for Digital Engineering, Norway); Carmen Guerrero (University Carlos III of Madrid, Spain); Adrián Gallego Sánchez (Universidad Carlos III de Madrid, Spain); Wint Yi Poe (Huawei Technologies – European Research Center, Germany); Yue Wang (Samsung Electronics, USA); Ahmed Mustafa Elmokashfi (Simula Research Laboratory, Norway)
Artificial Intelligence (AI) is widely applied in mobile and wireless networks to enhance network operation and service management. Advanced AI mechanisms often require high level of network service exposure in order to access data from as many network elements as possible and execute the AI recommended outcomes into the networks. However, in practice, it is not always feasible to expose the network services to 3rd parties or customers with AI ambitions. Considering that service assurance (SA) is a major area to which AI is applied, this paper describes how a closed-loop SA architecture is associated with the service exposure model in the 5G networks with network slicing. Then we investigate the impact and implication of service exposure on SA. Finally, a set of experiment results are provided to demonstrate the trade-off relationship between the AI ambition and the exposure level in SA.
Georgia D. Ntouni (Intracom Telecom, Greece); Thomas Merkle (Fraunhofer IAF, Germany); Eleftherios Loghis, Georgios Tzeranis, Vassilis Koratzinos, Nikolaos Skentos and Dimitrios S. Kritharidis (Intracom Telecom, Greece)
Antenna arrays implementing beamforming are required in Terahertz (THz) communications to combat the very high propagation losses. In this paper, real-time THz wireless transmission with beamforming at a carrier frequency of 300 GHz is demonstrated for the first time using a single-carrier FPGA-based modem and a uniform linear array with four antenna elements. Digital beamforming is implemented at the modem, which is a prototype based on a millimeter wave (mmWave) communication product. Moreover, suitable converter and signal conditioning boards as an interface to the THz frontends have been fabricated and utilized for the experimental setup. Apart from beamforming, THz specific imperfections of the transceivers as well as their impact on the communication performance have been experimentally explored.
Mikko Uitto (VTT Technical Research Centre of Finland Ltd, Finland); Antti Heikkinen (VTT Technical Research Centre of Finland, Finland)
In this paper, we describe the implemented 360 live video streaming architecture and evaluation setup running in the 5G test network (5GTN), which can be considered beneficial for live educational services and providing a fast situation awareness in medical sector. The concentration is on providing more global HTTP adaptive streaming (HAS) architecture using also CDN components rather than having only a point-to-point solution. With this setup, we evaluate the common media application format (CMAF) usage in HAS context in order to have reliable video transmission method for high bitrate 360 video as well as setting the end-to-end (e2e) latency as low as possible. By this, both uplink and downlink direction are investigated by using wired, wireless, or mobile connection. In addition, the experienced e2e latency is measured. The extensive placement of synchronized network measurement indicators from distinct network nodes help to identify the bottlenecks in mobile video delivery path and verify the CMAF-based usefulness in streaming optimization. 5G key performance indicators delay and throughput are key parameters in 360 live streaming, which can be useful in forming the overall, live view of an educational surgery or even from an accident scene. In such scenarios, mobile networking can be the easiest and often the only option for outputting the important information to relevant users, such as students or doctors.