Please use this identifier to cite or link to this item:
|Title:||DNN-based receiver for one-bit ADC distortion in OFDM system||Authors:||Gu, Ruixuan||Keywords:||Engineering::Electrical and electronic engineering::Wireless communication systems||Issue Date:||2022||Publisher:||Nanyang Technological University||Source:||Gu, R. (2022). DNN-based receiver for one-bit ADC distortion in OFDM system. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/158352||Abstract:||The Orthogonal frequency-division multiplexing (OFDM) system is extensively employed in the 5th generation wireless communication systems. In this dissertation, deep neural networks (DNNs) are used for channel estimation as well as signal detection instead of conventional ways. This dissertation starts from the one-to-one full-resolution OFDM system, based on which one-bit ADC quantization is added to form a low-resolution MIMO OFDM system. The simulation is separately conducted for the two parts. In the full-resolution OFDM system, the simulation procedure is separated into two steps. One is the offline stage, which aimed to minimize the difference between output recovered data and the transmitted data. The other one is the online stage, where the trained DNN is used to recover the online transmitted data. From the results, we can find the DNN methods can recover the transmitted data effectively and can handle channel distortion, which shows distinct robustness under inferior conditions compared with traditional methods. In the low-resolution MIMO OFDM system, another DNN is used to minimize the difference between the real channel vector ℎ and the estimated one ℎ#. It is interesting to find that as the quantity of antennas increases to a certain number, the SNR efficiency per antenna can reach an extremely high level, which is enough to ignore the bad influence of insufficient pilot length. This finding can help us to use fewer pilot symbols to obtain high antenna SNR, which can save more bandwidth and power to transmit useful data instead of long pilot symbols. In conclusion, DNN can be applied in the field of communications engineering. It can be utilized in channel estimation and symbol detection and shows good robustness. Although one-bit ADC sometimes may affect the accuracy of channel estimation and data detection, it is beneficial for us to find more appliable aspects to take advantage of its unique strengths.||URI:||https://hdl.handle.net/10356/158352||Schools:||School of Electrical and Electronic Engineering||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Theses|
Updated on Dec 4, 2023
Updated on Dec 4, 2023
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.