An Improved Channel Estimation Method for Cell-Free Uplink Systems With Scalar Quantizers

The channel estimation problem for the uplink of a cell-free multiple-input multiple-output (MIMO) system is studied. Standard approach for fronthaul quantization and channel estimation is reviewed first, and its limitation is discussed. To improve the channel estimation accuracy, it is proposed that, at each time instant, each radio unit (RU) quantizes a linear combination of the currently and previously received samples. To maximize the impact of linear combination, i.e., pre-processing, we tackle the problem of jointly optimizing the pilot sequences, pre-processing matrices and linear estimation filters with the goal of minimizing the sum of mean squared errors (MSEs) of channel gains. Numerical results validate the performance gains of the proposed channel estimation scheme with the optimized pre-processing matrices.