Keysight Technologies announced last Friday that it has collaborated with NVIDIA to create a complete design flow for training and validating neural receivers to be showcased at Mobile World Congress Barcelona 2024. This demonstration will be presented at the Keysight booth, Hall 5 Stand 5E12. Open RAN testbed powered by multi-user MIMO neural receivers.
While 5G will integrate artificial intelligence (AI) to power certain components of wireless networks, 6G will be the first generation of AI-native wireless technology. The main goal is to develop site-specific neural receivers that replace the entire human-designed physical layer receive chain. However, the data required to train these neural receivers is limited and validating their performance in end-to-end systems is a challenge. Before a neural receiver can be deployed in a commercial network, it must be properly trained, demonstrated to outperform traditional receivers in performance, and shown to robustly handle channel conditions in real-world networks.
This demo shows how Keysight’s solution enables the design and verification of neural receivers. Using the NVIDIA Sionna library to train neural receivers, raytracing channels enable the generation of site-specific training data, enabling the creation of digital twins of real-world systems. Therefore, neural receivers can be optimized for any intended environment.
Once training is complete, the neural receiver will be deployed in an open RAN testbed using Keysight equipment connected to LITEON Technology’s FlexFi commercial radio units. The Keysight PROPSIM channel emulator is used to emulate site-specific channels. This enables seamless import of raytraced channel impulse responses. A trained neural receiver then demodulates the signal. Measuring the block error rate of an end-to-end system provides insight into the performance of a neural receiver.
Giampaolo Tardioli, Vice President of Keysight 6G and Next Generation Technologies
By collaborating with industry leaders like NVIDIA, Keysight is helping provide the tools and insights needed to move artificial intelligence into the mainstream of wireless communications. By using simulated and measured data to train digital twins of open RAN networks and validate AI performance, we can provide complete end-to-end solutions for developing neural receivers and other AI components for 6G. We can provide a design environment. .
Ronnie Vasishta, Senior Vice President, NVIDIA Telecom
Software-defined RAN enables native integration of AI across the entire protocol stack, ultimately enabling 6G systems that are optimally adaptable to any environment. Keysight’s ability to generate and capture real-world data for training and evaluation, powered by the open source NVIDIA Sionna library for 6G physical layer research, is a key element in helping accelerate AI adoption in wireless networks.
