Search papers, labs, and topics across Lattice.
This paper introduces Lightweight PCGAE-Net, a novel architecture designed for efficient channel state information (CSI) prediction in 5G massive MIMO systems, addressing the challenges of deploying high-accuracy models with large parameter counts. By correcting architectural flaws in existing models, specifically through a redesigned attention mechanism and a Bottleneck AutoEncoder, the proposed model significantly reduces parameter size while enhancing prediction accuracy. The results demonstrate that PCGAE-Net, with only 8.54 million parameters, outperforms the CS3T-UNet baseline by up to 6.0 dB in challenging scenarios, showcasing its effectiveness for real-world applications.
Lightweight PCGAE-Net achieves superior CSI prediction accuracy with 58% fewer parameters than the leading model, revolutionizing efficiency in 5G systems.
Accurate channel state information (CSI) prediction is essential for proactive beamforming and resource management in 5G massive MIMO systems, yet the deployment of high-accuracy transformer-based predictors on base-station hardware remains challenging because the most capable models carry upwards of 30\,M parameters. This paper introduces Lightweight PCGAE-Net, which addresses the efficiency problem not by post-hoc compression but by correcting two architectural flaws in the current state of the art. The first is a sequential attention ordering bias: in CS3T-UNet, group-wise temporal attention (GTA) always operates on features that have already been transformed by cross-shaped spatial attention (CSA), distorting what temporal information GTA can capture. We remove this dependency by routing both attention modules to the same layer-normalized input and combining their independent outputs through a learned per-channel sigmoid CrossGate. The second flaw is an uncompressed bottleneck: applying full self-attention at the deepest encoder stage, where channel depth reaches $4C$, is quadratically expensive and carries redundant features. A Bottleneck AutoEncoder (BAE) with $1\times1$ convolutions halves this depth and uses an auxiliary reconstruction loss to prevent information collapse. Wrapping these components inside a shallower encoder-decoder with frequency-domain dimensionality reduction ($N_f\!=\!32$, $C\!=\!48$) produces a model with just 8.54\,M parameters -- 58\% fewer than the CS3T-UNet baseline -- that outperforms it by up to 3.26\,dB at 5\,km/h and 6.0\,dB at 9\,km/h in single-step prediction on QuaDriGa dataset.
