Quantum-inspired Tabu Search algorithm for antenna selection in massive MIMO systems
Quantum-inspired Tabu Search algorithm for antenna selection in massive MIMO systems
复制标题
DOI：
10.1109/wcnc.2018.8377099
复制
发表时间：
2018-04
期刊：
影响因子：
--
通讯作者：
Zaid Abdullah;C. Tsimenidis;M. Johnston
中科院分区：
文献类型：
--
作者：
Zaid Abdullah;C. Tsimenidis;M. Johnston
大规模多输入多输出（MIMO）系统通过在基站（BS）处使用大量天线单元可以显著提高系统性能和容量。然而，在BS处具有大量的射频（RF）链可能是昂贵的并且能量效率低。实现大规模MIMO系统的分集增益的一种方式是采用具有有限数量的RF链的大量天线。因此，可以应用天线选择技术来降低系统复杂度和硬件成本。在本文中，量子启发禁忌搜索（QTS）算法应用于大规模MIMO系统中的天线选择，并与两个著名的算法，即，经典禁忌搜索（CTS）算法和遗传算法（GA）进行了比较。QTS算法比CTS算法有很大的优势，因为它只需要找到最佳的旋转角度，使系统朝着更好的解决方案发展。相反，在CTS中，禁忌矩阵的维数是动态的，需要优化。此外，为了实现最大性能，当改变天线的数量或迭代次数时，这些维度需要重新配置，而在QTS中不会发生这样的问题。与CTS和GA相比，QTS算法在系统容量方面也显示出更好的结果。此外，经典和量子启发TS算法需要比GA低得多的复杂度。
Massive Multiple-Input Multiple-Output (MIMO) systems can significantly improve the system performance and capacity by using a large number of antenna elements at the base station (BS). However, having a massive number of radio-frequency (RF) chains at the BS can be costly and energy inefficient. One way to achieve the diversity gain of massive MIMO systems is to employ a massive number of antennas with limited number of RF chains. Thus, antenna selection techniques can be applied to reduce the system complexity and hardware cost. In this paper, a Quantum-inspired Tabu Search (QTS) algorithm is applied to antenna selection in Massive MIMO systems and compared with two well known algorithms; namely, a Classical Tabu Search (CTS) algorithm and a Genetic Algorithm (GA). The QTS algorithm has a great advantage over CTS, since it only requires finding the optimum rotation angle to evolve the system towards a better solution. In contrast, in CTS, the dimensions of the tabu matrix are dynamic and need to be optimized. Moreover, to achieve maximum performance, these dimensions need to be reconfigured when changing the number of antennas or the number of iterations, while no such a problem occurs in the QTS. The QTS algorithm also shows better results in terms of the system capacity compared to CTS and GA. Furthermore, the classical and quantum inspired TS algorithms require much lower complexity than the GA.