[ad_1]
With analysis on the know-how elements for the longer term 6G wi-fi communication customary in full swing, the chances of an AI-native air interface for 6G are additionally being explored. In collaboration with NVIDIA, Rohde & Schwarz takes one other step ahead and presents an enhancement of their latest hardware-in-the-loop demonstration of a neural receiver, exhibiting the achievable efficiency features when utilizing educated ML fashions in comparison with conventional sign processing, whereas for the primary time additionally optimizing the transmitter aspect.
At this 12 months’s Brooklyn 6G Summit, happening from October 31 to November 2 in Brooklyn, New York, guests can expertise the primary demonstration of how a neural receiver paired with a discovered customized modulation performs in a 5G NR uplink a number of enter a number of output (UL MIMO) situation. The setup combines premium take a look at options for sign technology and evaluation from Rohde & Schwarz and the NVIDIA Sionna GPU-accelerated open-source library for 5G and 6G link-level simulations. NVIDIA’s Sionna permits fast prototyping of advanced communications system architectures and gives native help to the combination of machine studying in 6G sign processing.
This demonstration builds on the profitable showcase on the Cellular World Congress 2023 in Barcelona, the place each firms demonstrated the idea of a neural receiver in a hardware-in-the-loop experiment involving two unbiased firms for the very first time. Academia, main analysis institutes, and key business consultants throughout the globe anticipate {that a} future 6G customary will use AI/ML for sign processing duties, corresponding to channel estimation, channel equalization, and demapping. This idea is known as a neural receiver. In the present day’s simulations recommend {that a} neural receiver will enhance total hyperlink efficiency and thus throughput in comparison with the present high-performance deterministic software program algorithms utilized in 5G NR whereas holding computational complexity at a manageable stage.
Going a step additional, the take a look at mattress has been prolonged to allow the verification of communication programs that apply AI/ML not solely within the receiver, but additionally within the transmitter. The demonstration at Brooklyn 6G Summit showcases the applying of discovered customized constellations, which researchers are investigating as a stepping stone in the direction of pilotless communication. As an alternative of counting on well-known, symmetric constellations corresponding to QPSK or QAM modulations, the constellation factors are decided by an end-to-end studying course of, which collectively optimizes the neural receiver and the constellation mapper of the transmitter whereas taking the light cell radio channel into consideration.
The take a look at mattress setup leverages the R&S SMW200A vector sign generator, which emulates a single person transmitting a 100 MHz (273 PRB) extensive sign within the uplink path at a service frequency of two.14 GHz in a MIMO 2×4 configuration. Fading and noise are utilized to the transmission to emulate sensible radio channel situations. The R&S MSR4 multi-purpose satellite tv for pc receiver captures the RF sign instantaneously with its unbiased 4x 200 MHz channels. The digitalized sign is streamed in real-time and transferred to the R&S Server-Primarily based Testing (SBT) framework together with R&S VSE vector sign explorer (VSE) micro-services, which performs synchronization and quick Fourier rework (FFT) calculation. The post-FFT information serves as enter for a neural receiver applied and educated utilizing NVIDIA Sionna. As a part of the demonstration, the educated neural receiver with customized constellation is in comparison with the classical idea of a linear minimal imply squared error (LMMSE) receiver structure, which applies conventional sign processing methods based mostly on deterministically developed software program algorithms. These high-performance algorithms are extensively adopted in present 4G and 5G mobile networks.
To confirm the efficiency of such an enhanced neural receiver, together with customized modulation, the capabilities of the R&S SMW200A vector sign generator had been prolonged for this demo to permit the person to freely configure every particular person constellation level by magnitude and section within the IQ area based mostly on the chosen QPSK or QAM modulation. The R&S FSW Sign and Spectrum Analyzer and the R&S Vector Sign Explorer (VSE) software program at the moment help customized modulation, too. This permits customers to evaluate transmitter efficiency by well-known key efficiency indicators such because the error vector magnitude (EVM) even when discovered customized constellations are used.
Rohde & Schwarz actively helps 6G analysis actions throughout Europe, Asia, and the US, whereas contributing to analysis initiatives, the work of business alliances, and collaborating with main analysis institutes and universities. The corporate’s take a look at and measurement experience and options assist pave the way in which for the subsequent technology of wi-fi communications, which is predicted to be commercially deployed round 2030.
[ad_2]