科研项目

Project description/goals

In this project, we investigate the distributed channel estimation in extremely large antenna array (ELAA) systems. In particular, for the two typical scenarios of spatially stationary and spatially non-stationary, distributed channel estimation algorithms with high precision, low complexity and low communication cost suitable for star network and chain network are designed respectively.

Importance/impact, challenges/pain points

The proposed distributed algorithms are suitable for the future distributed ELAA system, and perform close to the centralized algorithm, thereby improving the scalability of the network. However, to make the distributed algorithm approach the performance of the centralized algorithm, sufficient information exchange between nodes is required, which will bring huge information exchange overhead and computational complexity. The pain point of this project is how to design efficient distributed channel estimation algorithms under bandwidth and computing resources constraints.

Solution description

We propose two distributed channel estimation algorithms, namely the aggregation based distributed algorithm and the compressed sensing based algorithm. In the proposed algorithms, only a small amount of local information is required to be exchanged among local nodes to approach the centralized algorithm.

Key contribution/commercial implication

• In the low SNR (SNR=-20dB) case, when the average communication cost is no larger than 2% of a DMRS signal, the performance loss of the proposed schemes compared to the centralized algorithm does not exceed 0.5dB;
• In the high SNR (SNR=20dB) case, when the average communication cost is no larger than 20% of a DMRS signal, the performance loss of the proposed schemes compared to the centralized algorithm does not exceed 0.5dB;
• The computational complexity of the proposed algorithms are similar to the centralized scheme.

Next steps

Completed

Collaborators/partners

HUAWEI

Team/contributors

Tsung-Hui Chang, Yanqing Xu