Binary Neural Network Aided CSI Feedback in Massive MIMO System
This work addresses resource constraints in user equipment for wireless communication systems, representing an incremental improvement in efficiency.
The paper tackles the high memory and computation cost of deep learning-based CSI feedback in massive MIMO systems by introducing BCsiNet, a binarized neural network that reduces encoder memory usage by over 30x and speeds up inference by around 2x while maintaining comparable feedback performance to the original CsiNet.
In massive multiple-input multiple-output (MIMO) system, channel state information (CSI) is essential for the base station to achieve high performance gain. Recently, deep learning is widely used in CSI compression to fight against the growing feedback overhead brought by massive MIMO in frequency division duplexing system. However, applying neural network brings extra memory and computation cost, which is non-negligible especially for the resource limited user equipment (UE). In this paper, a novel binarization aided feedback network named BCsiNet is introduced. Moreover, BCsiNet variants are designed to boost the performance under customized training and inference schemes. Experiments shows that BCsiNet offers over 30$\times$ memory saving and around 2$\times$ inference acceleration for encoder at UE compared with CsiNet. Furthermore, the feedback performance of BCsiNet is comparable with original CsiNet. The key results can be reproduced with https://github.com/Kylin9511/BCsiNet.