Initial situation, problem to solve and motivation to carry out the R&D project – Connected autonomous vehicles are able to exchange sensor information (radar, optical, etc.), kinematic information and maneuver information to achieve cooperative joint decisions in difficult traffic situation forming an internet-of-things (IoT). This cooperation helps to improve traffic safety and approach the goal of zero accidents. Highly-reliable and low-latency 5G wireless vehicle-to-infrastructure (V2I) communication link shall enable the cooperation in a controlled manner connecting vehicles to a mobile-edge cloud computing center at the base-station location. V2I communication channels have challenging properties, they are strongly time-variant and exhibit non-stationary fading statistics due to the high mobility of vehicles. Hence, the design of a highly-reliable and low-latency wireless V2I communication link is a challenging task and new solutions are urgently needed.
Goals and level of innovation compared to the state-of-the-art – In MARCONI we will use the basic concept of massive multiple-input multiple-output (MIMO) systems where the base station is equipped with 30...100 antenna elements while the mobile station uses a single antenna. With this setup, it is possible to focus the transmit energy of the base station by coherent superposition at the location of the mobile station. Due to the sphere hardening effect in massive MIMO systems, the fading process can be practically removed enabling a constant transmission quality from the base station to the mobile station. In MARCONI, we will explore algorithms to enable these properties for the first time for highly time-variant and non-stationary V2I scenarios in both time-division duplex (TDD) and frequency division duplex (FDD) systems. To avoid shadowing we will devise algorithms for distributed massive MIMO antennas and implement the transceiver algorithms on a real-time software-defined radio testbed.
Expected results and findings – We will measure the radio wave propagation characteristics between multiple mobile nodes and the massive MIMO base station in vehicular scenarios. The measurement data is used to design a non-stationary geometry-based channel model for the numerical link level evaluation of massive MIMO transceiver algorithms for mobile nodes in vehicular scenarios. Time-variant subspace based channel estimation algorithms for massive MIMO systems utilizing hypothesis tests and compressed sensing concepts will enable the acquisition of high-quality channel state information. This data will be crucial for the massive MIMO precoding for time-variant channels. Here we will exploit the slow time-variation in the delay-Doppler domain to cope with mobile nodes providing a highly reliable and low-latency communication link. For rapid deployment in current network topologies we will investigate compressed channel state information feedback for operation in FDD systems. Thus, the already costly allocated frequency bands of the providers can be directly reused in the future. All algorithms in MARCONI will be tested in real-time on our software defined radio testbed with up to 96 antenna elements for frequencies up to 6 GHz.
- Starting date: September 2017
- Duration: 3 Years
- Funding: funded by the FFG (IKT der Zukunft - 5. Ausschreibung 2016)
- Coordination: AIT Austrian Institute of Technology
- Partner: Nokia Solutions and Networks, AVL List