ACE-6G: AI Communication Empowered Sematic Platform for 6G

Introduction

Pengcheng Laboratory and Beijing University of Posts and Telecommunications have made breakthroughs in nonlinear joint source and channel coding technology and the novel air interface technology of muti-user semantic communication and so on, and built ACE-6G: AI Communication Empowered Semantic Platform for 6G, which further improves the bandwidth efficiency by 2-3 times compared with traditional technologies, empowers 5G links to achieve 6G ubiquitous connection transmission capabilities, and provides strong support for the standardization of 6G key technologies.

Breaking Through Transmission Problems in Harsh Channel Environments and Achieving Ubiquitous Smart Interconnection

For the challenges of intelligent interconnection's requirements for network communication flexibility and adaptability, high-performance transmission in harsh channel environments, and high bandwidth requirements, the ACE-6G research team has proposed a semantic communication architecture based on a semantic knowledge base and designed a zero-shot multi-level feature transmission method enabled by semantic knowledge base, which realizes low-incremental knowledge base updates and reduces the communication service transmission latency by more than 90%. A breakthrough in joint source and channel coding technology based on nonlinear transform coding is made, which achieves that compared with 5G technology, the transmission bandwidth can be saved by50%, which improves the “cliff effect" of traditional methods in low signal-to-noise ratio transmission. A multi-user semantic communication prototype system based on new resource is developed, which incorporates the new dimension of semantic model domain resources and breaks through the new air interface technology of multi-user semantic communication. It can increase the spectrum utilization rate by at least 1 time compared with traditional technologies.

Focusing on 6G Intelligence and Communication Standardization to Empower Industry Applications

The achievement aims to break through the new paradigm of integration of 6G intelligence and communication, explores higher-level semantic information features, and develops semantic communication technology for 6G. The ACE-6G research team has published more than 50 high-level journal/conference papers on semantic communication, and realized the innovative application of semantic communication technology in scenarios such as low-altitude economy, intelligent networked unmanned vehicles, and satellite-to-ground direct connection. The ACE-6G research team has made a lead to found the semantic communication task group in IMT-2030 (6G) working group, completed the CCSA standardization research project, and released the technical white paper for semantic knowledge base, which has been supported by more than 10 well-known enterprises and universities, including China Mobile, Hua-wei and ZTE. It can promote the international standardization process of semantic communication for 6G. Moreover, the research team has submitted a total of 100invention patents, 32 of which have been authorized so far, and submitted 2 PCT patent applications. A new R&D institution has been jointly established with China Mobile and others to develop 6G basic theories and common key technologies, as well as explore industrial paths.

Opening New Paths for 6G and Inspiring Original Research on Semantic Communication

Facing the development trend of deep integration of 6G communication and intelligence, this achievement has proposed a semantic communication architecture based on semantic knowledge base,and realized the demonstration application of joint source-channel coding technology and multi-user semantic communication technology in low-altitude economy, high-definition video transmission, satellite communication and other scenarios. In view of the challenges of unmanned aerial vehicle communication with high requirements for airborne processing computing power and large consumption of network backhaul bandwidth, the incremental update and real-time transmission technology of semantic model are realized to reduce the demand for airborne computing power by more than 15% and the network backhaul bandwidth requirement by more than 70%, which no longer limits the development of low-altitude economy.

In order to meet the wireless transmission requirements of high-definition video in the intelligent network of unmanned vehicles, the research team has made a breakthrough in joint source-channel coding technology based on nonlinear transformation coding, which can save more than 50% of the transmission bandwidth of semantic communication compared with traditional wireless communication methods when the receiver achieves the same service quality in a harsh wireless channel environment with a packet loss rate of 30%. The semantic communication architecture and corresponding key technologies can build efficient transmission capacity, achieve high spectrum efficiency and better service experience for the 6G, and provide important support for the construction of network infrastructure for the digital economy.