Automotive applications of today rely on mobile network support for connectivity and have specific Quality of Service (QoS) requirements. Due to varying network conditions, it may not be always possible to fulfil these QoS requirements. In such cases, it is useful for the vehicle to obtain an in-advance notification about any upcoming changes in the available QoS so that it can take appropriate action, such as adapting or altogether stopping applications that can’t be operated in a safe manner under the predicted QoS conditions. This concept is called ‘Predictive QoS’.

Predictive QoS is being extensively discussed in industry organizations such as 5GAA as well as standards organizations such as 3GPP. Predictive QoS could be an important part of 3GPP release 17 and beyond. Therefore, it is very pertinent for the research community working in the field of 5G and beyond wireless technologies. The proposed workshop tries to bring together different aspects of predictive QoS, namely:

  • use-cases and requirements (from the automotive industry point of view)
  • implementation and operation (from an MNO or service provider point of view).

Additionally, the session aims to discuss the technical challenges associated with:

  • QoS prediction, e.g. accuracy of prediction, which parameters are most useful to be predicted, AI/ML models for prediction, the differences between long-term and short-term prediction
  • delivering the prediction, e.g. periodic vs event-triggered delivery, subscription vs relevance model for the receiver of the prediction
  • adaptation to the QoS prediction, e.g. reaction of the vehicular application to change in QoS, orchestration and/or arbitration of the adaptation.

With the increasing emphasis on achieving higher levels of automated driving, it is essential to develop connectivity concepts, such as predictive QoS, which will likely play an important role in enabling the corresponding automated driving applications. Hence, it is important for the research community to discuss such concepts presently. The outcome of such discussions will drive the timely development and standardization of these concepts.

Topics:

  • Machine Learning and AI for QoS prediction
  • Algorithms and techniques for timely notification of QoS change
  • Standardization requirements for QoS Prediction deployment
  • Architectures and algorithms for QoS prediction
  • Real-time QoS prediction requirements
  • Use Cases and scenarios for QoS prediction e.g., V2X applications, industrial applications
  • Application Adaptation based on QoS prediction
  • Privacy issues for data collection to enable QoS prediction notifications
  • Prototypes, and performance evaluation of QoS prediction for automated driving and other Industrial applications

Invited speakers:

“Artificial Intelligence-based QoS prediction in the AI4Mobile project“ – Prof. Slawomir Stanczak, Fraunhofer HHI

“Opportunities from Radio Access Network data for QoS prediction” – Dr. Alassane Samba, Orange Labs,

“QoS prediction in 5GCroCo project” – Dr. Dirk Hetzer, T-SYSTEMS

Papers:

“Effect of Spatial, Temporal and Network Features on Uplink and Downlink Throughput Prediction” – Alexandros Palaios (Ericsson Research, Germany); Christian Leonard Vielhaus (TU Dresden, Germany); Daniel Fabian Külzer (BMW Group Research and Technology, Germany); Philipp Geuer (Ericsson Research, Germany); Raja Sattiraju (University of Kaiserslautern, Germany); Jochen Fink (Technische Universität Berlin & Fraunhofer Heinrich Hertz Institute, Germany); Martin Kasparick (Fraunhofer Heinrich Hertz Institute & Technical University Berlin, Germany); Cara Watermann (Ericsson Research, Germany); Gerhard P. Fettweis (Technische Universität Dresden, Germany); Frank H.P. Fitzek (Technische Universität Dresden & ComNets – Communication Networks Group, Germany); Hans D. Schotten (University of Kaiserslautern, Germany); Slawomir Stanczak (Technische Universität Berlin & Fraunhofer Heinrich Hertz Institute, Germany)

“Using Transition Learning to Enhance Mobile-Controlled Handoff In Decentralized Future Networks” – Steven Platt (Universitat Pompeu Fabra, Spain); Berkay Demirel (Pompeu Fabra University, Spain); Miquel Oliver (Universitat Pompeu Fabra, Spain)

“One Step Further: Tunable and Explainable Throughput Prediction based on Large-scale Commercial Networks” – Roman Zhohov (Ericsson Research, Sweden); Alexandros Palaios and Philipp Geuer (Ericsson Research, Germany)

“LSTM-based QoS prediction for 5G-enabled Connected and Automated Mobility applications” – Sokratis Barmpounakis (University of Athens, Greece); Lina Magoula and Nikolaos Koursioumpas (National and Kapodistrian University of Athens, Greece); Ramin Khalili (Huawei Technologies, Germany); José Perdomo (Huawei Munich Research Center, Germany & Universitat Politècnica de València, Germany); Ramya Panthangi Manjunath (Huawei German Research Center, Germany)

“A Statistical Learning Framework for QoS Prediction in V2X” – Miguel Angel Gutierrez-Estevez (Huawei Technologies, Germany); Zoran Utkovski (Fraunhofer HHI, Germany); Apostolos Kousaridas (Huawei Technologies, Germany); Chan Zhou (Huawei European Research Center, Germany)

Workshop chairs:

  • APOSTOLOS KOUSARIDAS, Principal Research Engineer, Huawei Technologies, Germany
  • TIM LEINMUELLER, Senior Technical Manager, DENSO Automotive Deutschland GmbH, Germany