Page 310 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Partners

• Tsinghua University [7], Deep Transcend Co�, Ltd�

2�2 Benefits of use case

• The project promotes open-source, software-defined baseband architecture, encouraging
innovation and allowing more companies to enter the market. By integrating AI with
Wireless Baseband Processing (WBP), the project aims to accelerate technology iteration,
fostering industrial development.
• By reducing reliance on proprietary hardware and emphasizing open protocols, the
project allows small and medium enterprises (SMEs) and innovators to participate,
levelling the playing field in the telecommunications industry.
• The use of a RISC-V architecture optimized for AI-WBP synergy enhances energy efficiency,
contributing to more sustainable telecom infrastructure. The emphasis on open-source
and standardized designs encourages responsible production practices and sustainable
industry transformation.

2�3 Future Work

Future work on the open-source baseband architecture will focus on expanding protocol
support, advancing AI integration, and enhancing energy-efficient design. Efforts will be
made to establish robust collaborative platforms that link academia and industry to streamline
innovation. Opportunities to partner with leading telecommunications research bodies will be
explored to refine the architecture and address emerging industry needs. These initiatives aim
to develop a versatile platform that industry leaders, researchers, and policymakers can access
globally. Ultimately, the goal is to empower stakeholders with cutting-edge tools for strategic
decision-making, thereby fostering sustainable industry growth and innovation.

The roadmap focuses on three key areas: technical advancement, ecosystem development, and
scenario expansion. Technical advancements aim to expand protocols towards 6G, enhancing
AI-native capabilities such as federated learning frameworks, and achieving breakthroughs in
energy efficiency with near-memory computing. Ecosystem development will be highlighted at
ITU AI for Good, with the launch of expanded operator libraries and multi-functional toolchains
to create academia-industry collaboration platforms similar to PyTorch's ecosystem. Scenario
expansion targets emerging domains like aerial economy and communications in extreme
environments, transforming basebands from "closed black boxes" to "open innovation
platforms." This approach addresses 6G’s sustainability challenges by unifying performance,
flexibility, and economy through an open DSA architecture.

3 Use Case Requirements

• REQ-01: Open Protocol Implementation
It is critical that open protocols are developed and implemented to reduce reliance on
proprietary hardware and enhance interoperability.
• REQ-02: AI Integration with WBP
It is expected that AI capabilities will be integrated with Wireless Baseband Processing to
accelerate technology iteration and improve performance.
• REQ-03: Energy-Efficient Architecture
It is critical that a RISC-V domain-specific architecture is designed, focusing on energy
efficiency and adaptability to future communication standards.

274