Generalizing Deep Learning-Based CSI Feedback in Massive MIMO via ID-Photo-Inspired Preprocessing

TL;DR

This work tackles the generalization gap in DL-based CSI feedback for Massive MIMO by introducing UniversalNet, a preprocessing-driven framework that standardizes CSI inputs and enhances sparsity without updating backbone networks. It combines 2D DFT-based format standardization with an efficient eigenvector joint optimization to align channel statistics across environments. The approach yields substantial generalization gains on a ray-tracing WirelessAI2022 dataset, enabling high-quality CSI feedback even when training data is limited or drawn from unseen environments, and it preserves integration with existing DL models via lightweight preprocessing/postprocessing. The work demonstrates that modest preprocessing and eigenspace alignment can substantially reduce model mismatch in practical deployments, improving compression efficiency and downlink performance across diverse scenarios.

Abstract

Deep learning (DL)-based channel state information (CSI) feedback has shown great potential in improving spectrum efficiency in massive MIMO systems. However, DL models optimized for specific environments often experience performance degradation in others due to model mismatch. To overcome this barrier in the practical deployment, we propose UniversalNet, an ID-photo-inspired universal CSI feedback framework that enhances model generalizability by standardizing the input format across diverse data distributions. Specifically, UniversalNet employs a standardized input format to mitigate the influence of environmental variability, coupled with a lightweight sparsity-aligning operation in the transformed sparse domain and marginal control bits for original format recovery. This enables seamless integration with existing CSI feedback models, requiring minimal modifications in preprocessing and postprocessing without updating neural network weights. Furthermore, we propose an efficient eigenvector joint optimization method to enhance the sparsity of the precoding matrix by projecting the channel correlation into the eigenspace, thus improving the implicit CSI compression efficiency. Test results demonstrate that UniversalNet effectively improves generalization performance and ensures precise CSI feedback, even in scenarios with limited training diversity and previously unseen CSI environments.

Figures (4)

• Figure 1: Architecture comparison between the conventional DL-based solution (a) and proposed universal solution (b).
• Figure 2: Visualization of precoding matrix magnitudes for different environments using proposed preprocessing methods.
• Figure 3: CSI recovery comparison in the unseen environments at different feedback bits when the training set is collected from the given number of environments.
• Figure 4: Detailed CSI recovery comparison across a varying number of training environments, with feedback bits set around 45.