CSI Quantization for FDD Massive MIMO Communication

Abstract

We consider high-dimensional multiuser MIMO transmissions in Frequency Division Duplexing systems. For precoding, the frequency selective channel has to be measured, quantized and fed back to the base station by the users. In 5G New Radio (NR), a modular quantization approach has been applied for this, where first a low-dimensional subspace is identified for the whole frequency selective channel, and then subband channels are linearly mapped to this subspace and quantized. We analyze how the components in such a modular scheme contribute to the overall quantization distortion. Based on this analysis we improve the technology components in the modular approach. We compare the improved quantization scheme to the 5G NR standardized version by simulation in a scenario with a realistic spatial channel model. The improvements lead to a more than 25% improvement in spectral efficiency.

Modular CSI Quantization for FDD Massive MIMO Communication

“…5, each of the twelve combinations is simulated for several codebook sizes, and projection distortion is reported against the total number of bits. 1 Recalling that the projection distortion d p does not depend on orthogonalization, the same distortion arises for OWP and independent vector quantization IND, while SWP may have different d p . In Fig.…”

“…Here, a subtle effect comes into play. A given R has singular values 1 For clarity, the overlapping points from R B+B− are removed. The product of the TSODFT codebook over-sampling ratios in horizontal and vertical dimensions is ω = 2 n , n ∈ [2,9].…”

“…O. Tirkkonen is with Comnet, Aalto University, Finland (e-mail: olav.tirkkonen@aalto.fi). This work was funded in part by Huawei Technologies Co., Ltd. Part of the paper has been published in VTC Spring 2021 [1]. codebook size increases exponentially with the number of antennas, which prevents their application in massive MIMO networks, especially in frequency selective channels, where feedback is needed for each coherence bandwidth.…”

Low-Complexity Grassmannian Quantization Based on Binary Chirps

“…We simulate performance for m = 4, N = 16 and m = 5, N = 32. These are of practical interest for vector quantization of Frequency Division Duplexing massive MIMO communication, where first the antenna dimensionality is reduced by identifying an eigenspace, followed by vector quantization in a lower-dimensional subspace [8].…”

“…Subspaces of unit norm complex valued vectors modulo overall phase rotations are of interest in many information processing tasks, ranging from non-coherent wireless communication [1], [2], [3] and activity detection [4] to vector quantization of Channel State Information (CSI) [5], [6], [7], [8] and deep learning [9], while also being the fundamental objects of interest in finite dimensional quantum mechanics [10]. For communication [2], [3], quantization [5], [7], [8] and quantum coding [11] scenarios, codebook based approaches are essential.…”