Evaluations of the predistortion technique by neural network algorithm in MIMO-OFDM system using USRP

Autor: M Wisnu Gunawan, Naufal Ammar Priambodo, Melki Mario Gulo, Arifin Arifin, Yoedy Moegiharto, Hendy Briantoro
Jazyk: indonéština
Rok vydání: 2022
Předmět:
Zdroj: Jurnal Infotel, Vol 14, Iss 4, Pp 287-293 (2022)
Druh dokumentu: article
ISSN: 2085-3688
2460-0997
DOI: 10.20895/infotel.v14i4.825
Popis: MIMO OFDM is the key technology of 4G network system. MIMO-OFDM system enhances the spectrum efficiency and increases the capacity of the system. The implementation of USRP hardware to MIMO OFDM system has been attracted some researchers to conduct the experiments. So we conduct the experiments in a MIMO OFDM system that applies the predistortion technique. In this experiment, we evaluate performances of the predistortion technique by using the artificial neural network. USRP 2920 hardware which is supported by LabVIEW and Phyton software are used in this experiment. OFDM system uses 128 subcarriers to produce an OFDM symbol, and MIMO system uses 2 antennas at transmitter and receiver side. And no obstacles between Tx and Rx, or line of sight transmission scenarios. The performances of the predistortion technique using the artificial neural network algorithm are shown in symbol constellations or Error Vector Magnitude (EVM) at the receiver. And the texts or characters are used as the input of the system. From the experiment results can be seen that the distance between Tx and Rx affects the Error Vector Magnitude (EVM) and predistortion technique produces the Error vector magnitude (EVM) improvement. More shorter the distance between Tx and Rx can decrease distortions of the received signal, At the transmitter side, the performance of predistortion technique is shown as the linearization improvement of the non-linearity power amplifier. Therefore more wider the linear region of power amplifier results the decreasing in band distortion of transmitted signal, and can be seen as the Error Vector Magnitude (EVM) improvement.
Databáze: Directory of Open Access Journals
Externí odkaz: