Large-scale MIMO robust WMMSE precoder and deep learning design method thereof

Abstract

The invention discloses a large-scale MIMO robust WMMSE precoder and a deep learning design method thereof, and a base station uses statistical parameters of channel estimation values and channel estimation errors of user terminals to estimate the channel estimation errors of the user terminals through an iterative design or deep learning design method of the robust WMMSE precoder according to traversal and rates of all users or a traversal and rate lower bound maximization criterion. And calculating a precoding vector corresponding to each user terminal, and carrying out downlink robust WMMSE precoding transmission. The iterative design adopts a block coordinate descent method, and a statistical robust receiver, a weight parameter and a precoding vector are iteratively updated in sequence, so that the traversal and rate lower bound are maximized; according to the deep learning design method, based on a precoding vector structure determined by low-dimensional characteristic parameters, the low-dimensional characteristic parameters are calculated through a neural network, then precoding vectors are calculated through the structure, and downlink precoding achieves nearly optimal reachable and rate performance with low calculation complexity under various antenna configurations.

Description

technical field [0001] The present invention relates to downlink precoding of wireless communication, in particular to a massive MIMO robust WMMSE precoder and a deep learning design method thereof. Background technique [0002] Massive multiple-input-multiple-output (MIMO, massive multiple-input-multiple-output) can provide efficient communication services for a large number of users by configuring massive antennas at base stations (BS, base stations). The BS may pre-process the transmit signal through precoding to mitigate inter-user interference. [0003] Traditional precoders such as regularized zero-forcing (RZF, regularized zero-forcing) and signal-to-leakage-and-noise ratio (SLNR, signal-to-leakage-and-noise ratio) can achieve sub-optimal sum-rate performance; weighted least mean squares The error (WMMSE, weightedminimummean-square-error) precoder can maximize the sum rate, but since each iteration involves matrix inversion, the amount of computation is large, so the...

AI Technical Summary

Problems solved by technology

Random WMMSE counteracts the inaccuracy of CSI by iterating the channel samples multiple times, but each iteration involves matrix inversion, and its computational burden needs to be further reduced

Images