AI-Driven 5G Air Interface Design

The complexities of wireless communication are often obscured from consumers who enjoy the convenience of modern mobile devices. However, the underlying mechanics of wireless links can seem almost magical, relying on intricate data models. Therefore, managing these systems presents a ripe opportunity for Artificial Intelligence (AI) and Machine Learning (ML). The introduction of AI/ML for link management is outlined in 3GPP Release 18, and it’s anticipated to play a crucial role in the development of 6G and future technologies.

Qualcomm’s current demonstration highlights the application of AI/ML in two key areas: channel state feedback (CSF) and mmWave beam management. CSF is vital for determining modulation, coding, and various settings for wireless links, making its accuracy essential for optimal performance. Traditionally, networks acquire CSF through continuous device reporting. The new proposal in Release 18 suggests using AI/ML models on both devices and networks to estimate CSF, thus minimizing overhead and enhancing capacity while also reducing latency, as there's no need to wait for feedback. The demo illustrates how AI/ML estimates closely align with measured CSF.

The performance of 5G mmWave technology relies heavily on precise beamforming and beam-steering, which are based on statistical models. Utilizing AI/ML for beam management is an obvious choice, as demonstrated by improved outcomes in the demo.

The Metaverse Connection

No demonstration would be complete without a nod to the Metaverse! Qualcomm has long been engaged in AR/VR/XR research initiatives, well before the recent surge of interest in the Metaverse. Their demos encompass various aspects, including limitless XR with split rendering, latency optimizations, and mmWave-based tracking for a better integration of physical and digital experiences. Notably, a new modem-based API was developed to provide app developers with real-time radio condition data, enabling AR/VR and gaming applications to adapt swiftly to changing conditions.

Advancements in MIMO and mmWave Technologies

If you think the mmWave spectrum operates at a high frequency, consider Qualcomm’s exploration of the 145 GHz band, known as sub-Terahertz. This high frequency necessitates a special antenna technique called Lensed MIMO, promising Terabit/sec speeds. The prototype demonstrated achieves speeds of up to 100 Gbps using over 100 GHz of spectrum, potentially revolutionizing connectivity.

Another demo showcased the utilization of the 13 GHz spectrum in the upper mid-band. This system, using antennas of similar sizes, can deliver five times the bandwidth while maintaining the same coverage, significantly enhancing capacity compared to the current 3.5 GHz systems. The band itself ranges from 7 to 24 GHz.

Before these futuristic concepts become reality in 5G Advanced and 6G, Qualcomm also displayed numerous technologies aimed at immediate applications, such as Sub-Band Full-Duplex (SBFD) and mmWave evolution. SBFD, introduced in Release 18, aims to enhance latency, particularly for industrial applications, with the potential to evolve into single-frequency full-duplex in 6G. The demo effectively illustrates the performance gains possible with SBFD, a technology Qualcomm has been developing for some time.

The mmWave enhancement demos include mobility improvements for applications requiring ultra-low latency, shared spectrum deployment, and a mmWave network planning tool that uses 3D street models for better coverage predictions.

Enhancing Positioning Capabilities

Accurate positioning has become increasingly crucial, particularly for Industrial IoT applications where robots must be effectively managed in factories. Various prototypes and simulations demonstrate positioning through mid-band, 5G mmWave, 60 GHz mmWave, and the Reduced Capacity (RedCap) category defined in Release 17. These utilize techniques such as RF fingerprinting, AI/ML, and angle of arrival/departure for enhanced accuracy.

Improvements in Automotive Safety

The importance of radars in autonomous vehicles is well-known, yet they can still have blind spots. A collaborative radar demo illustrated how vehicles can mitigate these blind spots through cloud connectivity or vehicle-to-vehicle (V2V) communication. This demo emphasizes challenges that often go unnoticed until highlighted.

Other Cellular V2X demonstrations utilize roadside units to provide precise positioning in areas with limited GPS coverage, such as urban canyons, as well as enhancing safety applications.

IoT Innovations and Industrial Applications

A variety of demos related to IoT were also presented. The wide-area IoT demos simulated the capacity of RedCap devices, showcasing the network's ability to support a dense network of lower-complexity devices while extending IoT coverage through mesh connectivity. Industrial IoT demos illustrated robust, high-reliability connectivity in factories, adhering to O-RAN Alliance specifications and utilizing techniques like Coordinated MultiPoint (CoMP).

Additionally, Qualcomm showcased how power savings in networks can be achieved, aligning with 3GPP’s initiative for greener networks—an essential goal for Release 18.

For your convenience, Qualcomm has made videos of all their demos available on YouTube.