Submitted by Automa??o e Estat?stica (sst@bczm.ufrn.br) on 2016-12-15T18:09:01Z No. of bitstreams: 1 TalesViniciusRodriguesDeOliveiraCamara_DISSERT.pdf: 3587261 bytes, checksum: ddffd176ad921074a281b50e904134f5 (MD5) / Approved for entry into archive by Arlan Eloi Leite Silva (eloihistoriador@yahoo.com.br) on 2016-12-28T19:00:56Z (GMT) No. of bitstreams: 1 TalesViniciusRodriguesDeOliveiraCamara_DISSERT.pdf: 3587261 bytes, checksum: ddffd176ad921074a281b50e904134f5 (MD5) / Made available in DSpace on 2016-12-28T19:00:57Z (GMT). No. of bitstreams: 1 TalesViniciusRodriguesDeOliveiraCamara_DISSERT.pdf: 3587261 bytes, checksum: ddffd176ad921074a281b50e904134f5 (MD5) Previous issue date: 2016-04-25 / T?cnicas de Classifica??o Autom?tica de Modula??o (AMC) t?m sido utilizadas por sistemas modernos de comunica??o para otimizar o uso do espectro e com isso aumen- tar as taxas de transmiss?o de dados. No processo de AMC, v?rias arquiteturas podem ser utilizadas para retirar informa??o e avaliar caracter?sticas do sinal modulado em um canal. Uma grande parte dessas arquiteturas s?o constru?das utilizando como base a ci- cloestacionariedade. A an?lise cicloestacion?ria ? realizada por meio das ferramentas: Fun??o de Autocorrela??o C?clica (CAF) e Fun??o Densidade Espectral C?clica (SCD). Esta ultima particularmente, ? utilizada para observar as caracter?sticas cicloestacion?rias de diferentes sinais, as quais s?o chamadas de assinaturas. Embora tenha v?rias aplica- ??es bem sucedidas no ?mbito de AMC, a cicloestacionariedade possui restri??es pois a CAF e SCD s?o limitadas ? an?lise estat?stica de segunda ordem, devido ao uso da correla??o com cerne de sua express?o. Com o objetivo de generalizar a avalia??o da cicloestacionariedade sobre infinitos momentos estat?sticos de um sinal, surgem Fun??o de Autocorrentropia C?clica (CCAF) e a Fun??o Densidade Espectral de Correntropia C?clica (CCSD). Tais fun??es s?o fundamentadas no c?lculo da correntropia. Neste tra- balho a CCSD ser? investigada quanto capacidade de gerar assinaturas para diferentes modula??es e seu potencial de uso em AMC ser? avaliado. / The steady growth in the use of wireless communication systems has contributed to finding new ways to exploit the maximum capacity of use spectrum. In this context, cognitive radios appear as an appropriate option able to offer an efficient use of the channel, ensuring greater bandwidth to users. In the scenario of cognitive radios, cyclostationary analysis techniques have shown to be quite effective in extracting features that can be used in the spectrum sensing. Such features called cyclostationary signatures are generated by the spectral correlation density function (SCD) and can be directly associated with the type of modulation used on the channel. Architectures for spectrum sensing using SCD has good performed when used in AWGN channels. However, recent studies show that the tool doesn?t have a good performance in the extraction of signal characteristics contaminated with impulsive noise (Outlier), because it is limited to second order statistical analysis. In order to generalize the SCD cyclostationary analysis for endless statistical moments, arise the function correntropy cyclic spectral density (CCSD) This work proposes a spectrum sensing architecture using CCSD, which is applied to the extraction cyclostationary features from digital modulations: ASK, FSK, BPSK, QPSK and MSK. The sensing architecture proposed is evaluated in various parameters: different sensing thresholds, change of SNR levels of a AWGN channel, different kernel sizes (s) from CCSD and extraction of cyclostationary features from modulations contaminated with noise impulsive. The results of this study demonstrate the effectiveness of the proposed architecture.