Adaptive initial beam search for sparse millimeter wave channels

Abstract

Narrow beamforming solutions for millimeter wave cellular networks require mobile and base stations to exhaustively search all angular space. This scanning is mandatory in order to detect the unique optimum beamforming and combining vectors prior to control and data signaling. However exhaustive search yields large computational complexity and increases beam discovery time and control plane latency. Hence this paper presents some novel adaptive search algorithms to accelerate initial beam acquisition in sparse millimeter wave channels in non-line of sight environments. Namely the beam search is formulated as an optimization problem, and fast search procedures are presented based upon Luus Jaakola, divide and conquer, and Tabu direct pattern methods. The proposed algorithms deliver significant performance improvements versus existing beam search solutions, i.e., 25–99%.