Low-Altitude Economy Benefits from AI-Enhanced Wireless Communications Systems.

Research demonstrates the application of large language models (LLMs) to enhance near-field communications within the emerging low-altitude economy. LLMs address complexities in signal processing and user distinction, optimising spectrum efficiency for communication with unmanned aerial vehicles by intelligently managing beam energy direction and multi-user precoding.

The burgeoning low-altitude economy, encompassing unmanned aerial vehicles and related services, demands increasingly sophisticated communication infrastructure. Researchers are now exploring the potential of integrating large language models (LLMs), a type of artificial intelligence adept at processing complex data, with near-field communication systems. Near-field communication operates over short distances, utilising extremely large multiple-input multiple-output (XL-MIMO) antenna arrays to focus energy precisely on target devices, offering enhanced spectral efficiency. However, implementing this technology presents challenges in signal processing and user identification. Zhuo Xu, Tianyue Zheng, and Linglong Dai, Fellow of the IEEE, address these issues in their article, “Empowering Near-Field Communications in Low-Altitude Economy with LLM: Fundamentals, Potentials, Solutions, and Future Directions”, presenting a scheme utilising LLMs to distinguish between near and far-field users and optimise multi-user precoding.

The convergence of Large Language Models (LLMs) and near-field Extremely Large-Scale Multiple-Input Multiple-Output (XL-MIMO) systems offers a potentially significant advancement for communication networks supporting the developing Low-Altitude Economy (LAE). This research examines how LLMs mitigate challenges inherent in near-field communications, with specific application to areas such as drone delivery, aerial surveillance, and smart city infrastructure, establishing a basis for utilising artificial intelligence in future network development. The central concept involves employing the predictive and analytical capabilities of LLMs to optimise critical elements of physical layer communication, thereby creating more intelligent and resilient networks.

Near-field XL-MIMO systems, which utilise a large number of antennas at both the transmitter and receiver to improve data throughput and reliability, present considerable potential but demand sophisticated signal processing. LLMs can predict channel state information (CSI), which describes how a radio signal propagates from transmitter to receiver, reducing the need for overhead signalling and improving spectral efficiency, reliability, and overall system capacity. This represents a departure from conventional physical layer communication approaches, offering a route to more intelligent and robust networks.

LLMs exhibit a capacity for multi-task learning, concurrently managing tasks such as user identification and precoding design, a capability particularly valuable given the variable demands of LAE applications. Precoding involves manipulating the transmitted signal to optimise its quality at the receiver. By accurately predicting channel conditions and optimising beamforming, a technique that focuses radio signals in specific directions, the system maximises the utilisation of available radio spectrum, while robust interference management and precise CSI contribute to more dependable communication links.

The proposed LLM-based scheme offers advantages beyond simply improving signal transmission; it establishes a foundation for leveraging artificial intelligence to construct the next generation of communication networks for the low-altitude economy, potentially enabling more efficient and reliable operation of emerging aerial technologies and services.