Crowd-sourcing and Model-driven based Online Dictionary Learning for FDD Massive MIMO Systems

Abstract

In this paper, we propose a crowd-sourcing and model-driven based online dictionary learning for FDD massive MIMO systems. This method is applied to address the problem of uplink and downlink channel estimation in FDD massive MIMO systems and overcome the shortcomings of traditional offline dictionary learning based channel model. Offline dictionary learning must be carried out at the cell deployment stage and need to collect a large number of channel measurements. We introduce the idea of crowd-sourcing and distribute the task of data collection to the users in the cell, hence the learning can be carried out even when the cell is working and the burden of data collection is alleviated. Furthermore, offline dictionary cannot adopt to the change of cell environment, while online dictionary learning can keep up-to-dated with new crowd-sourcing information. In addition, with the assistance of prior information, we develop a model-driven based method to accelerate the convergence speed of the learning process. The combination of crowd-sourcing and model-driven ensures the online dictionary learning higher efficiency and stronger robustness, as is proven in the simulation results.