How to Verify the Performance of a 6G Neural Receiver

Signal Generator
+ Signal Generator

Employing AI to optimize 6G neural receiver designs

Verifying the performance of a 6G neural receiver requires site-specific training data generation. Yet, limited data is available, and validating the receiver’s performance in end-to-end systems is challenging. Once the algorithm design is complete, designing a 6G neural receiver using artificial intelligence (AI) is a multiple-step process. Before deploying neural receivers in commercial networks, engineers must ensure that the receivers are well-trained, outperform traditional receivers, and handle the channel conditions of real-world networks robustly. AI integration in 6G focuses on two processes: channel estimation and channel state feedback.

If engineers do not understand channel behavior and fail to compensate for its anomalies in real time, 6G performance will fall consistently short of expectations. Design engineers need a solution to train neural receivers using software-generated labeled data. After generating the data, they need to validate the neural network’s performance when integrated into a wireless system. Then, they can emulate and integrate different channel conditions into the system. A channel emulator is necessary to import channel models from external tools or use existing model data. This approach enables engineers to create a digital twin of various channel conditions and compare the simulation results with a real-world system.

6G AI neural receiver design test solution

6G AI neural receiver design test solution

Verifying receiver functionality in 6G systems requires accurate channel estimation. The Keysight solution trains a neural receiver using labeled data generated by Keysight PathWave System Design. The system optimizes the training data to be site-specific, and the data updates to accommodate different scenarios. When the training is complete, the Keysight equipment generates and transmits new 5G waveforms to the neural receiver through a live Open RAN network, complete with a commercial ORAN radio unit. The receiver can then demodulate the signal using AI and machine learning algorithms from the previous training. Once the system processes the signal, test engineers can measure the bit error rate / block error rate for the end-to-end system to provide insights into the neural receiver’s performance. Finally, the Keysight PROPSIM channel emulator can import channel models from external ray-tracing tools. The data then serves as the digital twin of a channel to compare simulation results with a real-world system.

See demo of 6G AI neural receiver testing

Explore products in our 6G AI neural receiver design test solution

Discover resources and insights

Additional resources for verifying 6G neural receiver design

Related use cases

contact us logo

Get in touch with one of our experts

Need help finding the right solution for you?