Learning Optimal Beam Directions in Next-generation Wireless Networks: A Fixed-Budget Stochastic Bandit Approach
Millimeter Millimeter Wave (mmWave) communication and Intelligent Reflecting Surfaces (IRS) are promising technologies to improve throughput in 5G and beyond 5G networks. However, any misalignment between the Transmitter (Tx) and the Receiver (Rx) beams significantly reduces their throughput. Hence one needs to learn the beam directions that offer good throughput between Tx and Rx. We propose algorithms to learn the beams that offer the best throughput using the Multi-Armed Bandits (MAB) framework. In wireless networks, scheduling protocol design and implementation can be simpler if the channel estimation is completed in a fixed duration before the data transmission can start. Thus, we focus on the Fixed-Budget Best Arm Identification (FB-BAI) MAB setup to identify the best beam. As it is desirable to use fewer pilots for learning, we exploit the unimodal structure of the throughput (as a function beam index) to improve the error performance of the FB-BAI algorithms. We develop an algorithm named 'Unimodal Bandit for Best Beam (UB3)' and establish its superior performance analytically and experimentally compared to the state-of-the-art.
Manjesh Kumar Hanawal received a BTech degree in ECE from NIT, Bhopal, in 2004, an M.S. degree in ECE from the Indian Institute of Science, Bangalore, India, in 2009, and the Ph.D. degree from INRIA, Sophia Antipolis and University of Avignon, France, in 2013. After two years of postdoc at Boston University, he joined Industrial Engineering and Operations Research at the Indian Institute of Technology Bombay, Mumbai, India, where he is an associate professor now. During 2004-2007 he was with CAIR, DRDO, working on various security-related projects. His research interests include communication networks, machine learning, and cybersecurity. He is a recipient of Inspire Faculty Award from DST and the Early Career Research Award from SERB. He has received several research grants like MATRICS from SERB and Indo-French Collaborative Scientific Research Programme from CEFIPRA. His work received the best paper award (honorable mention) at COMSNETS 2018.