Low-Complexity ML and Near-ML Detector Designs for Generalized Spatial Modulation

• Main Authors: Yueh-Lun Chang, 張岳綸
• Other Authors: Chih-Peng Li
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/gbfdd5

碩士 === 國立中山大學 === 通訊工程研究所 === 107 === Generalized spatial modulation (GSM) is a novel multiple-input-multiple-out (MIMO) technique, in which only several transmit antennas are activated in each time slot. Although the maximum likelihood (ML) detector is able to achieve the optimal performance, exhaustive search leads to computational complexity that is difficult to handle. In this paper, a search method called A-star algorithm is adopted, which uses a special heuristic function to directly remove some obviously poor paths. It can have lower complexity without losing performance. However, to determine the value of each node, we use the cost function of codebook-assisted hard decision (CAHD), which is a tree search system. CAHD is a low-complexity detector by setting the number of paths to be reserved for each layer. But when a certain signal is calculated, the other signals and noise are regarded as interference. This has a low complexity, but there is an error floor at high SNR. Our proposed algorithm effectively solves error floor of CAHD because we use the A-star search algorithm. This simulation results show that we can maintain the performance of ML, and the complexity is still lower than ML. On the other hand, we use different weights at different SNR to propose performance near-ML, but the complexity can be more significantly reduced.