GTI Secure Federated Learning Challenge (Research topic area supported by GSMA)
We are pleased to announce that the GTI Secure Federated Learning Challenge has ended successfully and the winner is the Universiti Malaya. Congratulations!
Traditional centralized AI models face significant privacy and scalability issues, particularly when handling sensitive or distributed data (e.g., medical records, smart-city sensor data). Federated Learning (FL) addresses these challenges by enabling decentralized model training across edge devices—raw data remains local, and only encrypted model updates are transmitted. However, large-scale adoption of FL in 5G/6G networks is hindered by three key barriers: security vulnerabilities (e.g., model poisoning, data leakage), device heterogeneity (varying hardware/software capabilities), and communication bottlenecks (high latency/bandwidth constraints in dense networks).
To develop a novel, secure, and scalable Federated Learning (FL) architecture specifically designed for deployment in 5G/6G telecom networks, overcoming the aforementioned barriers and enabling reliable, privacy-preserving AI applications.
- Applications open: [Thursday 31st July 2025]
- How to Register: Submit your information and PDF proposal at the bottom of this page
- Applications close: [Tuesday 19th August 2025 11.59pm Beijing time]
- GTI Judging: [To Friday 22nd August 2025]
- Official Announcing: [Results will be announced after GTI judging, the winner will be invited to attend the 2nd GTI Forum on Digital Intelligence, Hongkong, China, Sept 9-10]
- Prize: Award through contract with China Mobile of funding up to RMB 1.6 million for research and development laboratory equipment and environment and program to commence from contract award across 2 years from time of contract award. Research outputs to be jointly owned by winning organization and China Mobile.
- Contact: admin@gtigroup.org
Potential Applications:
- Smart-city (e.g., traffic management with distributed sensor data)
- Healthcare Systems (e.g., collaborative medical image analysis without sharing raw patient data)
- Autonomous Systems (e.g., connected vehicles with decentralized decision-making)
- 6G Networks (e.g., self-optimizing network management via federated models)
Application Requirements for Participants
Participants are required to submit a detailed proposal covering the following:
Research Qualifications:
Demonstrated expertise in AI (especially FL), 5G/6G network technologies, and cybersecurity (e.g., publications, past projects, team credentials).
Laboratory Construction Plan:
A clear blueprint for building a dedicated FL test environment, including:
- Laboratory space layout (to accommodate 5G/6G infrastructure and edge computing nodes).
- Equipment (must include 5G Stand-Alone (SA) core, edge servers, and secure communication tools).
- Technical capabilities (e.g., support for FL model training, privacy protection mechanisms, and 5G/6G network simulation).
Support Capabilities:
Explanation of how the proposed laboratory will support FL research in 5G/6G networks, such as testing secure model aggregation, optimizing communication efficiency, and validating real-world application scenarios.
Proposals will be assessed by GTI based on:
Laboratory Readiness:
- Availability of dedicated laboratory space.
- Access to 5G SA core equipment and related infrastructure.
Technical Capabilities:
- Quality of datasets and simulator tools (e.g., NS-3, MATLAB) for FL and 5G/6G testing.
- Expertise in AI, 5G/6G networks, and security (evidenced by team profiles and past work).
Feasibility of the Plan:
- Practicality of the laboratory construction blueprint.
- Alignment with the goal of developing a secure, scalable FL architecture for 5G/6G.
Incentives and Implementation Plan
Funding for Winners:
After evaluation, 1 winning team will receive dedicated funding from GTI up to GB Pounds £165k (RMB 1.6 million) to implement their proposed laboratory construction plan, including equipment procurement, infrastructure setup, and initial research activities.
Official Announcement:
The winning team will be officially announced at the GTI Forum in Hong Kong in September, followed by the signing of a cooperation agreement with China Mobile(Hongkong) Innovation Research Institute (covering funding details and collaboration objectives).
The challenge aims to establish a world-class FL laboratory, accelerate the development of secure and scalable FL architectures for 5G/6G, and foster industry collaboration between to drive real-world adoption of federated AI.
Selected Frequently Asked Questions (FAQs)
1. What is the expected duration and phasing of the research program ?
- From Co-operation agreement signing with China Mobile – 2 years total duration of program
- Phase 1 months 1-6: Federated network integration
- Phase 2 months 7-12 : Security framework, analysis of algorithms
- Phase 3 months 13-24: Evaluation of algorithm, algorithm optimisation, preparation of findings for publication
2. Can the lab and the research be located anywhere in the world - or does the lab need to be located in China ?
From GTI’s perspective the lab and the research can be located anywhere in the world.
Applicants should be aware of any local applicable regulations in their own jurisdiction.
3. After completion of the research who will own (have the intellectual property rights to) the research output ?
The intellectual property rights to the output of the research will be jointly owned by China Mobile and the winning organization.
4. Will GTI provide any data sets , or will there be any pre-determined partners providing data sets ?
GTI will not be providing data sets and there are no pre-determined partners to provide data sets.
Applicants should include in their proposal the proposed existing or future source of data sets for the program.
