Ray-of-Arrival Passing for Indirect Beam Training in Cooperative Millimeter Wave MIMO Networks

This paper is concerned with the channel estimation problem in multi-cell Millimeter (mmWave) wireless systems. We develop a novel Ray-of-Arrival Passing for Indirect (RAPID) Beam Training framework, in which a network consisting of multiple BS are able to work cooperatively to estimate jointly the UE channels. To achieve this aim, we consider the spatial geometry of the mmWave environment and transform conventional angular domain beamforming concepts into the more euclidean, Ray-based domain. Leveraging this model, we then consider the conditional probabilities that pilot signals are received in each direction, given that the deployment of each BS is known to the network. Simulation results show that RAPID is able to improve the average estimation of the network and significantly increase the rate of poorer quality links. Furthermore, we also show that, when a coverage rate threshold is considered, RAPID is able to improve greatly the probability that multiple link options will be available to a user at any given time.