Improving the Efficiency of MIMO Simulations in ns-3
Improving the Efficiency of MIMO Simulations in ns-3
复制标题
DOI：
10.1145/3592149.3592167
复制
发表时间：
2023-05
期刊：
影响因子：
--
通讯作者：
Matteo Pagin;S. Lagén;Biljana Bojović;Michele Polese;M. Zorzi
中科院分区：
文献类型：
--
作者：
Matteo Pagin;S. Lagén;Biljana Bojović;Michele Polese;M. Zorzi
信道建模是无线技术和网络的设计和评估的基本任务，在实际的原型设计、商业产品开发和真实的部署之前。当前和未来移动的网络的最新趋势（包括大天线系统、大规模部署和高频带）需要复杂的信道模型来精确仿真毫米波（mmWave）和太赫兹（THz）频带中的大规模MIMO（m-MIMO）。为了解决复杂度/准确度的权衡，3GPP（TR 38.901）已经定义了空间信道模型，这已经被证明是ns-3中当前系统级仿真的主要瓶颈。在本文中，我们专注于提高大规模MIMO系统级仿真的信道建模效率。扩展是在两个方向发展。首先，我们通过允许使用本征库进行更有效的矩阵代数运算，以及其他优化和更模块化的代码结构，提高了ns-3中当前3GPP TR 38.901实现代码的效率。其次，我们提出了一种新的面向性能的MIMO信道模型，以降低复杂度，作为适合毫米波/太赫兹频段的替代模型，并根据3GPP TR 38.901模型对其进行校准。仿真结果表明，适当的校准新引入的模型为各种场景和信道条件，并表现出有效的减少仿真时间（高达16倍相比，以前的基线）由于各种建议的改进。
Channel modeling is a fundamental task for the design and evaluation of wireless technologies and networks, before actual prototyping, commercial product development and real deployments. The recent trends of current and future mobile networks, which include large antenna systems, massive deployments, and high-frequency bands, require complex channel models for the accurate simulation of massive MIMO (m-MIMO) in millimeter wave (mmWave) and Terahertz (THz) bands. To address the complexity/accuracy trade-off, a spatial channel model has been defined by 3GPP (TR 38.901), which has been shown to be the main bottleneck of current system-level simulations in ns-3. In this paper, we focus on improving the channel modeling efficiency for large-scale MIMO system-level simulations. Extensions are developed in two directions. First, we improve the efficiency of the current 3GPP TR 38.901 implementation code in ns-3, by allowing the use of the Eigen library for more efficient matrix algebra operations, among other optimizations and a more modular code structure. Second, we propose a new performance-oriented MIMO channel model for reduced complexity, as an alternative model suitable for mmWave/THz bands, and calibrate it against the 3GPP TR 38.901 model. Simulation results demonstrate the proper calibration of the newly introduced model for various scenarios and channel conditions, and exhibit an effective reduction of the simulation time (up to 16 times compared to the previous baseline) thanks to the various proposed improvements.