Co-funded by the Federal Ministry for Economic Affairs and Energy (BMWi)
Duration: March 2020 - March 2023
Powerful and flexible mobile networks are a key technology for numerous branches of industry. Many applications, for example in the fields of automated vehicles or mobile robotics, are highly dynamic and mobile. At the same time, they require extremely reliable communication with low latency. An important challenge is to ensure the required quality of service (e.g. a reliable data rate, no data loss, no delays) for each application. The wireless transmission of information is by nature subject to interference, which means that successful real-time transmission in classic, reactive mobile radio systems cannot be guaranteed. Forward planning of network resources and prediction of relevant quality of service parameters are therefore essential, especially for safety-critical functions. Given the immense complexity of the overall network, methods of artificial intelligence (AI) offer a promising approach here.
The project AI4Mobile aims at developing methods that enable a robust prediction of the quality of service at high mobility using data from the entire mobile network. In combination with other data (application, environment and network data) the predicted quality of service information will be used to develop AI mechanisms for real-time network optimization. AI methods are particularly promising for resource management. In order to make use of existing knowledge in the field of mobile communications, the possibility of supplementing AI approaches with area-specific context information and of integrating expert knowledge into the design of learning processes will be investigated. This will enable a hybrid approach of classical algorithm development and artificial intelligence. The project investigates to what extent this approach can increase the security and robustness of mobile networks.
The AI4Mobile solution can be used to predict critical quality of service parameters (e.g. data rate, delay) of a complete communication link between sender and receiver. This allows the implementation of proactive resource optimization and networking in all parts of the mobile infrastructure. An early detection of critical states ensures that the performance of the systems is maintained even if undefined states and faults occur. The security and robustness of the mobile radio network is significantly increased, which is essential for dynamic applications such as automated driving or mobile robotics. The project thus contributes to consolidating Germany's technological leadership, especially in the important automotive and industry 4.0 sectors.
In addition to leading the project, the HHI contributes to AI4Mobile by developing AI-based methods for quality-of-service (QoS) prediction and for the configuration of essential system parameters in mobile networks. One focus is on the prediction and corresponding configuration of suitable air interface parameters for cellular and direct (V2V) communication. The developed concepts will first be evaluated by means of simulations. Additionally, an implementation of demonstrators in the application areas of vehicle and factory communication (especially mobile robotics) is planned. This can build on preliminary work from the 5GNetMobil project.