Apply Deep Learning to Investigation the Detection Efficiency of 5G System over 3-D alph-lamda-mu Fading Channel

• Main Authors: LIN, KAO-TON, 林國棟
• Other Authors: CHEN, YUNG-TSUNG
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/4f87g5

博士 === 大葉大學 === 電機工程學系 === 106 === A novel design for degrading or decreasing the system performance due to the spatially channel correlated phenomenon by using of the artificial intelligence (AI) will be proposed in this project. Based on the AI the deep learning algorithms will be applied to solve the problems mentioned above. Frequently, the issues of correlated channel are solved by applying the data scrambling over forward link and avoiding the inter-symbol interference (ISI) repeatedly, respectively. Firstly, in the project plans apply the algorithms of Convolution Neural Network, CNN)、Auto-encode (AE)、Belief Propagation (BP) to located at the receive terminal. Next, the de-channel correlation (DCC) technique with deep learning is deployed at the transmitter. It is predicated that there will meet many challenges in the way to establish the 3-D alph-lamda-mu correlated channel model. The beamforming technology used in the antenna pattern can be adopted to collect the data for feeding to the input of deep learning. The simulation and analyzed results will compare to the one that obtains from the tradition MIMO radio systems with correlated channels. The system performance is going to be optimized by the work out for the 3-D alph-lamda-mu massive MIMO communication in order to decrease the degradation of spatially correlated channel.