On The Way To 6G: Rohde & Schwarz Presents AI/ML-Based Neural Receivers With NVIDIA At The MWC2023

While research on the technology components for the future 6G mobile communications standard is already in full swing, the possibility of an AI-native air interface for 6G is also coming into focus. Rohde & Schwarz, in collaboration with NVIDIA, goes a step beyond pure simulation and lays the foundations for the incorporation of artificial intelligence and machine learning (AI/ML) into future 6G technology. At the MWC in Barcelona, ​​the companies will present the industry's first hardware-in-the-loop demonstration of a neural receiver, showing the performance gains that can be achieved by using trained ML models over traditional signal processing.

At this year's Mobile World Congress, visitors can expect the first demonstration of the performance of a neural receiver in a 5G NR uplink, multi-user, multiple-input, multiple-output (MU-MIMO) scenario - a blueprint for a possible 6G -Physical layer. The setup combines high-end test solutions for signal generation and analysis from Rohde & Schwarz with the open-source NVIDIA Sionna GPU-accelerated library for link-level simulation.

The concept of a neural receiver envisages that the signal processing blocks of the physical layer of a radio communication system are replaced by trained machine learning models. Scientists, leading research institutes and industry experts around the world expect that a future 6G standard will use AI/ML for signal processing tasks such as channel estimation, channel equalization and demapping. Today's simulations suggest that such a neural receiver can increase link quality and further boost throughput compared to the already powerful deterministic software algorithms used in 5G NR.

Appropriate data sets are of course required to train such machine learning models. However, the required data sets are often only available to a limited extent or not at all. At the current state of early 6G research, using metrological tools to generate different datasets with different signal configurations represents a viable alternative to train the ML models for signal processing tasks.

In the AI/ML-based neural receiver setup shown at the Rohde & Schwarz booth, the R&S SMW200A vector signal generator emulates two users transmitting an 80 MHz wide signal in the uplink direction with a MIMO 2x2 signal configuration. Each user is independently faded and noise is applied to simulate realistic radio channel conditions. The R&S MSR4 acts as the receiverUniversal satellite receiver that captures the signal transmitted with a carrier frequency of 3 GHz via its four phase-coherent reception channels. It then makes the data available to a server via its real-time streaming interface. There the signal is pre-processed using the R&S Server-Based Testing (SBT) framework including the micro-services of the R&S VSE Vector Signal Explorer (VSE) software. The VSE signal analysis software synchronizes the signal and performs a Fast Fourier Transform (FFT). This FFT-transformed dataset serves as input to the neural receiver realized with NVIDIA Sionna.

NVIDIA Sionna is an open-source GPU-accelerated library for link-level simulations. It enables rapid prototyping of complex communication system architectures and provides native support for integrating machine learning into 6G signal processing.

During the demonstration, the trained neural receiver is compared to the classical concept of an LMMSE (Linear Minimum Mean Squared Error) receiver architecture, which uses traditional signal processing techniques based on deterministically developed software algorithms. These already very powerful algorithms are widely used in today's 4G and 5G cellular networks.

Andreas Pauly, Executive Vice President Test and Measurement at Rohde & Schwarz, explains: “Signal processing in mobile communications using machine learning algorithms is a highly topical topic in the industry that is often the subject of controversial discussion among colleagues. We're excited to be working with a partner like NVIDIA on this experimental field. Our cooperation will enable researchers and industry experts to validate their models in a data-driven manner using our leading test solutions for signal generation and analysis and to test them in a hardware-in-the-loop experiment.”

Ronnie Vasishta, Senior Vice President of Telecommunications at NVIDIA, comments, “Trained ML models unlock significant performance potential compared to traditional signal processing. This hardware-in-the-loop demonstration of a neural receiver from Rohde & Schwarz and NVIDIA is a milestone for the industry and demonstrates the value of AI and machine learning for 6G technology.”

Rohde & Schwarz actively supports 6G research in Europe, Asia and the USA. The company is involved in research projects and industry alliances and works with leading research institutes and universities. The metrological know-how and the solutions from Rohde & Schwarz are paving the way for the next generation of mobile communications, 6G, which is expected to be commercially launched around the year 2030.

Rohde & Schwarz is showing the AI/ML-based trained neural receiver at Mobile World Congress 2023 at Fira Gran Via in Barcelona, ​​Hall 5, Stand 5A80. Interested visitors are cordially invited to convince themselves of the performance gain of the neural receiver and to discuss the details and the overall concept with the experts from Rohde & Schwarz and NVIDIA.

Further information on Rohde & Schwarz at the MWC Barcelona can be found at: www.rohde-schwarz.com/mwc

About Rohde & Schwarz
With its leading solutions in the areas of Test & Measurement, Technology Systems and Networks & Cybersecurity, the Rohde & Schwarz technology group is one of the pioneers of a secure and networked world. Founded more than 85 years ago, the group is a reliable partner for its customers from business and the public sector around the world. As of June 30, 2022, the number of employees worldwide was around 13,000. The independent group generated sales of 2.53 billion euros in the 2021/2022 financial year (July to June). Company headquarters are in Munich.