XAI-Driven Client Selection for Federated Learning in Scalable 6G Network Slicing

XAI improves trustworthiness in federated learning: The paper demonstrates that using explainable AI (XAI) methods to select which devices participate in network training makes the system more transparent and trustworthy for users, addressing the "black-box" problem where AI decisions cannot be understood or verified.

Privacy protection without sacrificing performance: By using federated learning with XAI-guided client selection, the approach allows network slices to keep their data private while still training AI models collaboratively—the simulations showed faster convergence time and lower computational costs compared to standard methods.

Scalability for next-generation networks: The method was designed specifically for 6G network environments where vast amounts of user data would normally create communication bottlenecks; the XAI approach enables the system to operate at scale while maintaining both speed and efficiency.

Responsible automation in critical infrastructure: The paper addresses a key responsible AI concern: ensuring that AI systems managing communication networks operate with transparency and reliability, which is essential when deploying automated systems in live production environments that millions of users depend on.

Handles real-world data challenges: The approach works effectively in "non-IID scenarios" (where data is unevenly distributed across devices), which reflects real-world conditions and makes the solution more practically applicable than methods that assume idealized data distributions.