This research introduces a federated learning approach for telecommunications networks that enables collaborative AI training while preserving data privacy.

• Compressed tiny language models (TinyLLMs) enable efficient deployment on mobile network equipment
• Federated learning allows multiple network cells to train models together without sharing sensitive data
• Achieves 90%+ accuracy in predicting network features with significantly reduced model size
• Creates self-optimizing networks that automatically adjust configurations based on predicted requirements

For telecommunications engineering, this breakthrough enables truly Autonomous Networks that can self-optimize, self-repair, and self-protect while maintaining data security and operational efficiency.

Efficient Federated Learning Tiny Language Models for Mobile Network Feature Prediction