Learning Optimal Beam Directions in Next-Generation Wireless Networks - A Fixed-Budget Stochastic Bandit Approach

Explore a technical lecture on optimizing beam directions in next-generation wireless networks through a fixed-budget stochastic bandit approach. Delve into the challenges of Millimeter Wave communication and Intelligent Reflecting Surfaces in 5G and beyond, focusing on solving beam misalignment issues between transmitters and receivers. Learn about the Multi-Armed Bandits framework and its application in developing efficient beam direction algorithms, particularly the Fixed-Budget Best Arm Identification setup. Discover the innovative 'Unimodal Bandit for Best Beam (UB3)' algorithm and understand its superior performance in improving throughput with minimal pilot usage. Benefit from the expertise of Prof. Manjesh Kumar Hanawal, an associate professor at IIT Bombay, whose extensive experience spans communication networks, machine learning, and cybersecurity, and whose work has earned multiple prestigious awards and research grants.