Keysight Technologies, Inc. has partnered with Qualcomm Technologies, Inc. to showcase an innovative machine learning (ML)-based Channel State Information (CSI) compression technique designed to improve link adaptation efficiency in advanced Multiple-Input Multiple-Output (MIMO) systems. The companies presented the demonstration at Mobile World Congress (MWC) Barcelona 2026, highlighting how artificial intelligence can enhance the performance of next-generation wireless networks.

During a controlled laboratory validation, the ML-driven CSI feedback approach delivered more than 40 percent improvement in downlink throughput compared with the conventional 3GPP eType II CSI reporting method in four-layer (rank-4) operation. This significant gain underscores the potential of AI-enabled network optimization for evolving 5G-Advanced deployments.

AI Authority Trend: Keysight and Qualcomm Accelerate RF Digital Twins for 5G-Advanced and 6G Massive MIMO

As wireless networks continue to evolve, 5G-Advanced technologies increasingly rely on a higher number of antennas and wider spectrum channels. Consequently, Channel State Information plays a critical role in helping networks determine how to steer beams and select optimal transmission parameters. However, as CSI becomes more detailed, it also generates greater uplink signaling overhead. This challenge becomes particularly pronounced in higher-layer MIMO scenarios, where networks must balance improved performance with efficient signaling.

To address this challenge, Keysight and Qualcomm Technologies conducted extensive testing using a mobile test platform powered by a Qualcomm 5G Modem-RF system, combined with Keysight’s advanced network emulation solutions. The testing environment allowed engineers to evaluate performance under repeatable and controlled conditions.

Within these constraints, the ML-based CSI compression method demonstrated superior efficiency when compared to the standardized eType II reporting technique. More specifically, it delivered a more compact yet accurate representation of channel conditions while preserving the critical data required for precise beamforming and link adaptation decisions. As a result, the solution enables scalable advanced MIMO configurations and supports ongoing industry efforts to integrate AI-native capabilities into the physical layer of future wireless networks, including emerging 6G systems.

AI Authority Trend: Keysight, Samsung and NVIDIA Showcase Unified AI-RAN Validation at MWC 2026

Tingfang Ji, Vice President of Engineering and Head of 6G Research at Qualcomm Technologies, Inc., said:
“Efficient CSI feedback is fundamental as antenna dimensions and spatial layers increase. This demonstration shows how Qualcomm’s advanced machine learning algorithms can significantly improve downlink performance over conventional eType II CSI reporting, supporting continued 5G-Advanced innovation and laying the foundation for future AI-native 6G systems.”

Lucas Hansen, Vice President and General Manager of the Wireless Test Group at Keysight, also emphasized the role of AI in next-generation connectivity. He noted that integrating AI-driven capabilities at the physical layer will play a vital role in enabling scalable wireless infrastructure.

Lucas Hansen, Vice President and General Manager, Wireless Test Group, Keysight, said: “AI-native enhancements at the physical layer are essential to scaling next-generation wireless networks. By combining Qualcomm’s modem test platform with our AI-enabled base station emulation solutions, we are enabling rigorous evaluation of ML-based CSI compression and its impact on advanced MIMO performance and 6G research directions.”

Overall, the collaboration highlights how AI and machine learning are rapidly becoming integral to wireless innovation, particularly as the industry moves toward more advanced 5G capabilities and begins laying the technological groundwork for future 6G networks.

AI Authority Trend: Keysight Launches Scale-Up Validation Portfolio for AI Data Center Interconnects

To share your insights, please write to us at info@intentamplify.com