Analysis and optimizations of PMI and rank selection algorithms for 5G NR
Multiple-Input Multiple-Output (MIMO) is crucial for enhancing spectral efficiency, channel capacity, coverage, and robustness. However, it requires significant computations to determine a precoding matrix for transmitted data streams. In closed-loop MIMO, as adopted in 3GPP 5G NR, these computation...
| Autores: | , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) |
| Repositorio: | r-CTTC. Repositorio Institucional Producción Científica del Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) |
| OAI Identifier: | oai:cttc.fundanetsuite.com:p8690 |
| Acceso en línea: | https://cttc.fundanetsuite.com/Publicaciones/ProdCientif/PublicacionFrw.aspx?id=8690 |
| Access Level: | acceso abierto |
| Palabra clave: | Ns-3 5G-LENA PMI and RI selection CSI feedback Codebook-based precoding type-I System-level simulation 5G NR |
| Sumario: | Multiple-Input Multiple-Output (MIMO) is crucial for enhancing spectral efficiency, channel capacity, coverage, and robustness. However, it requires significant computations to determine a precoding matrix for transmitted data streams. In closed-loop MIMO, as adopted in 3GPP 5G NR, these computations occur on the user side. To avoid transmitting large matrices, 3GPP defined codebooks with pre-defined precoding matrices indexed by the Precoding Matrix Indicator (PMI). The User Equipment (UE) selects a PMI and a Rank Indicator (RI) to report to the Next Generation Node Base (gNB) as part of the Channel State Information (CSI) feedback. PMI/RI selection can be done via exhaustive search or more efficient techniques, which are crucial for real UE implementations due to their impact on computational complexity and energy consumption. This paper analyzes various PMI/RI selection techniques using the open-source ns-3 5G-LENA simulator. We have implemented state-of-the-art techniques in the system-level simulator and carried out extensive simulation campaigns. Also, we propose new PMI/RI selection methods by focusing on performance versus computational complexity tradeoffs. Our proposed techniques show a superior simulation speedup (3.71x to 1.119x) with minimal throughput degradation (3% to 3.3%) compared to exhaustive search, depending on sub-band downsampling settings. Other state-of-the-art techniques implemented exhibit greater throughput losses (up to 8.3%) for a lower speedup (up to 3.54x) or similar losses with smaller speedups and potential slowdowns. |
|---|