Multi-User mmWave Beam and Rate Adaptation via Combinatorial Satisficing Bandits
This work introduces SAT-CTS, an algorithm for multi-user mmWave systems that learns optimal beam and data rate assignments without knowing channel information. It uses a 'satisficing' approach, aiming to meet a minimum throughput goal rather than just maximizing it. The method provides the first regret bounds for this type of learning in combinatorial semi-bandits.