AI-powered 6G receiver – Hardware-in-the-loop testing of an AI receiver integrating DPoD

Join Andreas Oeldemann from Rohde & Schwarz and Dani Korpi from Nokia Bell Labs as they demonstrate how to evaluate the performance of HybridDeepRX, Nokia Bell Labs’ advanced AI receiver, in hardware-in-the-loop experiments. The companies partnered to develop a testbed for evaluating the performance of HybridDeepRx in the lab.

In mobile communications, the signal you send often gets distorted. This leads to a high error vector magnitude (EVM), which traditional receivers struggle to decode correctly. A technology addressing this problem is HybridDeepRx: an AI receiver that enables the base station to process the received wireless signals, even when those signals are distorted and noisy.

The test bed is based on the R&S®SMW200A vector signal generator and FSWX signal and spectrum analyzer from Rohde & Schwarz. We use the generator to produce 5G uplink signals that mimic a mobile device and its channel-emulation to add fading and noise. To emulate non-linear distortion, a power-amplifier model is applied to the signal before it goes through the wireless channel. On the receiver side, the FSWX in combination with the VSE (Vector Signal Explorer) software records the impaired signal and analyzes it leveraging neural network inference using the HybridDeepRx AI receiver from Nokia Bell Labs demonstrating the benefit of the integrated DPoD that leads to enhance coverage for future 6G networks.