Name
Abstract | ||
|---|---|---|
High overhead of sharing and feedback and high computational complexity are common problems in multi-cell processing. In this paper, a novel framework for bidirectional signal transformation between space and frequency domains of massive MIMO channels is proposed to reduce system processing overhead and complexity. We design new space and frequency features and build the framework by two off-line trained neural networks (NN). Moreover, the uniqueness of spatial features is proved. Average errors of uni- and bi-directional transformation are 7.6% and 73%. When applying the framework to inter-cell interference coordination (ICIC), the system and edge throughput are both increased compared to the traditional scheme with low information sharing overhead. |
Year | DOI | Venue |
|---|---|---|
2021 | 10.1109/GLOBECOM46510.2021.9685595 | 2021 IEEE GLOBAL COMMUNICATIONS CONFERENCE (GLOBECOM) |
Keywords | DocType | ISSN |
Bidirectional transformation framework, space domain, neural networks, massive MIMO, 3D channel model | Conference | 2334-0983 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
6 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Lei Zhu | 1 | 0 | 0.68 |
| Guoliang Gao | 2 | 0 | 0.68 |
| Kai Li | 3 | 0 | 2.03 |
| Yang Yang | 4 | 612 | 174.82 |
| Liantao Wu | 5 | 0 | 1.69 |
| Fanglei Sun | 6 | 0 | 2.70 |