China Mobile Research Institute's Wireless and Terminal Technology Research Department Director, Li Nan, was recently invited to attend the Fifth Electromagnetic Spectrum Academic Conference's "Low-Altitude Smart Connectivity and Embodied Collaboration Forum." There, he delivered a keynote speech titled "Building an Electromagnetic Environment and Network Empowerment for the Low-Altitude Economy," addressing the pain points facing wireless communication networks amid the rapid development of the low-altitude economy. He systematically introduced four key technical matrices developed by China Mobile Research Institute: sparse sample multi-level intelligent augmentation for electromagnetic space, a three-dimensional network quality assessment system, an end-to-end integrated optimization handover strategy, and network-aware path planning for drones.
Leveraging these technical matrices, China Mobile Research Institute has initiated the construction of a low-altitude electromagnetic data augmentation platform to support the low-altitude economy's growth. Li Nan pointed out that the low-altitude economy, as a national strategic emerging industry, relies on a high-performance 3D communication network as a critical foundation. Adopting a problem-oriented and demand-driven approach, China Mobile has delved into frontline operations to identify industry pain points and key challenges, summarizing four primary challenges for low-altitude networks: coverage, evaluation, handover, and path planning. First, how to expand limited low-altitude flight test data into full three-dimensional coverage, solving the "incomplete coverage" issue. Second, how to establish a new evaluation system to accurately measure the quality of 3D low-altitude networks, addressing the "difficult evaluation" problem. Third, how to manage frequent handovers caused by multi-layer sidelobe coverage and long-distance line-of-sight propagation, preventing "service interruptions." Fourth, how to integrate communication quality prior information into route path planning, solving the "uninspected route" challenge.
Facing these challenges, China Mobile leverages its network advantages to deeply innovate network technologies, building a network-based, intelligence-driven capability for electromagnetic environment construction and network empowerment tailored to the low-altitude economy. This effort reinforces the network foundation for the industry's development. Regarding the sparse sample multi-level intelligent augmentation technology for electromagnetic space, Li Nan aptly described it as "data augmentation." To address the pain point that limited flight test quality cannot represent the full 3D airspace coverage quality—as low-altitude coverage expands to 3D space—China Mobile employs multi-cell joint modeling to design a CMOE hybrid expert model. By introducing an adaptive feature mapping layer to enhance model generalization, it achieves data augmentation from "line" to "area." This technology can save at least 40% in flight test costs. In live network tests across 66 cells in three provinces, the average accuracy of data augmentation exceeds 90%, effectively solving the challenge of perceiving full low-altitude coverage.
In terms of the 3D network quality assessment system, traditional ground network evaluation metrics, being "old rulers," fail to accurately measure 3D low-altitude network quality. To address this, China Mobile has built a three-layer assessment architecture—"business scenarios, network requirements, and evaluation indicators"—comprising over 20 indicators across five dimensions, along with quantitative scoring rules. This system precisely characterizes the multi-dimensional digital profile of low-altitude network quality. It can evaluate network quality from multiple angles and clearly identify specific optimization directions. For example, in one city evaluation, it accurately pinpointed issues with 4.9G network coverage and user experience. Based on this achievement, China Mobile has led the initiation of an industry standard project in CCSA TC7.
Regarding the end-to-end integrated optimization handover strategy, to tackle performance degradation caused by multi-layer sidelobe coverage and long-distance line-of-sight propagation in co-frequency air-ground networking scenarios, China Mobile has developed an end-to-end integrated optimization handover strategy. This ensures stable network experience for low-altitude services. By researching a line-of-sight, long-distance missing neighbor cell identification algorithm and proposing a deep reinforcement learning-based hybrid experience replay global optimization algorithm, it quickly finds the globally optimal CIO parameters. Laboratory tests show that this strategy reduces frequent handover occurrences by 82%, effectively resolving issues like service stuttering and data flow interruptions.
For network-aware path planning of drones, traditional path planning lacks prior communication information, suffers from low learning efficiency in avoidance strategies, and has poor generalization. To address this, China Mobile has designed a network-aware priority experience replay and poor-quality risk detection mechanism, enabling precise avoidance of poor-quality areas. By setting stable communication, poor-quality avoidance, and low energy consumption as multi-objective goals, it constructs an efficient reinforcement learning planning framework, achieving accurate avoidance of poor-quality zones. Verified through live network test data, this technology improves the drone flight path's poor-quality area avoidance rate by approximately 96% and enhances the average SINR quality by over 35%, effectively ensuring service quality.
Li Nan introduced that China Mobile, centered on an "AI-powered low-altitude data intelligent augmentation engine," has initiated the construction of a full-process low-altitude electromagnetic data intelligent augmentation platform. By integrating multi-source data and innovating model generalization, it connects the "collection, augmentation, and application" full-chain loop, breaking through the limitations of traditional flight test data to achieve efficient prediction and intelligent modeling of the low-altitude electromagnetic space. This platform has been trialed and validated in multiple provinces and cities in live networks, supporting the thriving development of the low-altitude economy. Li Nan stated that looking ahead, China Mobile is willing to collaborate with industry partners, converging innovation and industrial chains, to deeply cultivate low-altitude network technology innovation, intelligently empower new quality development, and contribute greater strength to building a stronger network nation and digital China.
Comments