Mamba-assisted Neural Network Enhances OFDM Channel Estimation Efficiency

Global: Mamba-assisted Neural Network Enhances OFDM Channel Estimation Efficiency

Researchers from several institutions, including Dianxin Luan and co‑authors, have introduced a neural network framework that leverages the Mamba architecture together with self‑attention mechanisms to improve channel estimation for orthogonal frequency‑division multiplexing (OFDM) waveforms. The paper was submitted to arXiv on 23 January 2026.

Technical Approach

The proposed system, termed MambaNet, integrates a customized Mamba backbone designed to process the large number of subcarriers typical of modern OFDM configurations. By embedding a self‑attention layer, the network captures long‑range dependencies across subcarriers, enabling more accurate reconstruction of channel state information while maintaining a low computational footprint.

Bidirectional Selective Scan

Unlike conventional Mamba implementations, the authors incorporate a bidirectional selective scan that processes subcarrier data in both forward and reverse directions. This design addresses the non‑causal nature of channel gains across subcarriers, allowing the model to consider information from both preceding and succeeding frequencies during estimation.

Complexity Advantages

The authors report that MambaNet exhibits lower space complexity compared with transformer‑based neural networks commonly used for similar tasks. By reducing the number of tunable parameters, the framework achieves a more compact representation without sacrificing estimation accuracy.

Simulation Evaluation

Simulation experiments were conducted using the 3GPP TS 36.101 channel model. Results indicate that MambaNet outperforms several baseline neural‑network solutions, delivering improved mean‑square error performance while employing fewer trainable parameters.

Potential Impact

If adopted in practical wireless systems, the reduced computational demands of MambaNet could facilitate real‑time channel estimation on resource‑constrained devices, supporting the deployment of high‑density OFDM networks with extensive subcarrier counts.

This report is based on information from arXiv, licensed under Academic Preprint / Open Access. Based on the abstract of the research paper. Full text available via ArXiv.