Banca de QUALIFICAÇÃO: JOSE LENIVAL GOMES DE FRANCA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : JOSE LENIVAL GOMES DE FRANCA
DATE: 02/10/2020
TIME: 09:00
LOCAL: Google Meet
TITLE:

Spectrum Sensing of Channels with Non-Orthogonal Multiple Access in the Power Domain Using Cyclostationary Analysis

KEY WORDS:

Non-Orthogonal Multiple Access, Cognitive Radio, Spectrum Sensing, Cyclestationary Analysis, HDFT, Sparse Fourier Transform

PAGES: 30
BIG AREA: Engenharias
AREA: Engenharia Elétrica
SUMMARY:

The fifth generation of mobile technologies (5G) sets standards that try to anticipate and meet the expectations of demand for mobile communications for the 2020s. 5G requirements, such as high transmission rates, low latency, and a large number of devices connected simultaneously, seek to be overcome by new multiple access technologies that optimize the use of both the spectrum already available and new bands that are yet to be explored. Two of these multiple access technologies, Radio Cognitive (CR) and Non-Orthogonal Multiple Access (NOMA), have individually gained special attention from academia and industry. In the NOMA in the power domain (PD-NOMA), several user equipments (UE) with predetermined power levels occupy the same spectrum at the same time. In this context, techniques based on cyclostationary analysis can indicate not only the occupation of the spectrum but also extract characteristics of the modulations that allow us to distinguish them. The spectral sensing can be done using second-order cyclostationary analysis, from the extraction of cyclostationary characteristics (cyclostationary signatures) performed with the Cyclic Autocorrelation Functions (CAF) and Cyclic Spectral Density (SCD), both dependent on the calculation of Transformed Fourier and presenting high computational cost. This work proposes an architecture for extracting cyclostationary signatures from PD-NOMA signals. PD-NOMA signals come from the UE uplink process using QPSK, 16QAM, and 64QAM modulations according to the predetermined power level. The calculation of CAF and SCD is done using FFT, HDFT (Hopping Discrete Fourier Transform), and Sparse Fourier Transform, and the performances of the three techniques are compared. The architecture's extension to distinguish users according to the proposed modulations will be evaluated through the use of higher-order cyclostationarities. The architecture is used in the spectrum sensing stage of a CR and uses neural networks as a classifier to determine the users' presence. The network is trained by providing signature information corresponding to each possible set of users in a scenario where the channel degrades the signal with Additive White Gaussian Noise (AWGN). The architecture evaluation is made in a scenario with the variation of the SNR levels of an AWGN channel.

BANKING MEMBERS:
Presidente - 1543191 - LUIZ FELIPE DE QUEIROZ SILVEIRA
Externo à Instituição - FABRÍCIO BRAGA SOARES DE CARVALHO - UFPB
Externo à Instituição - BRUNO BARBOSA ALBERT - UFCG