Made available in DSpace on 2016-04-04T18:31:36Z (GMT). No. of bitstreams: 1 Rodrigo Frandsen da Silva.pdf: 3782928 bytes, checksum: cd013447a70916926978a2ddc8e42dbb (MD5) Previous issue date: 2012-12-19 / Pontif?cia Universidade Cat?lica de Campinas / In this dissertation we evaluate a new technique that performs optical encryption of signals travelling through transparent optical networks (TON). It is, thus, intended to prevent eavesdroppers to capture and retrieve optical signals. The technique consists in dividing an optical signal into several spectral slices and applying different attenuation and delays to each of them. After this process the signal is again multiplexed and transmitted through the considered TON. At this point the signal is ideally unintelligible to any receiver who does not know the encryption key, i.e. the set of utilized attenuations and delays. To evaluate the strength of such key, we measure the bit error rate (Bit Error Rate, BER) of the encoded signal, BERC. Generally speaking, the higher BERC, the lower is the chance of the encrypted signal being decoded by an eavesdropper. At the authorized receiver side, signal is again divided into spectral slices which are subjected to a set of attenuations and delays that are complementary to the ones utilized as the encryption key. All slices are again multiplexed and, as a result, at the output of the decoder the original encoder input signal is ideally reconstructed. The quality of the decoded signal is measured by evaluating the BER of the decoded signal, BERd, which should be low enough to allow proper reception of the signal transmitted by the TON. Simulations with the software VPITransmissionMaker, VPIPhotonics Company Inc., were used to investigate the performance of the technique for different gains and delays. For the operation of spectral slicing, the profiles of the utilized filters were adjusted to represent the ones of state-of-the-art off-the-shelf equipment. Results indicate that BERC may reach values of up to 42% and 32% for non-return to zero (NRZ) and differential quadrature phase shift keying (DQPSK) signals, respectively. In both of these cases it was possible to find results of BERd lower than 10-12, after properly adjusting the spacing between the filters; this was observed even for propagation over amplified standard fiber links with lengths as long as 400 km. To the best of our knowledge, this is the first time that such analyses are presented. / Neste trabalho avaliamos uma t?cnica para realizar a criptografia totalmente ?ptica em redes ?pticas transparentes. Dessa maneira, pretende-se impedir que um intruso consiga capturar e analisar o sinal ?ptico. A t?cnica consiste em dividir um sinal ?ptico em diversas fatias espectrais e aplicar diferentes atenua??es e atrasos a cada uma delas. Ap?s este processo o sinal ? novamente multiplexado e transmitido por uma rede ?ptica transparente. Neste ponto o sinal est? idealmente inintelig?vel para qualquer receptor que n?o conhe?a a chave criptogr?fica (conjunto de atenua??es e atrasos) utilizada. Para avaliar a for?a da chave criptogr?fica, mede-se a taxa de erros de bit (Bit Error Rate, BER) do sinal codificado, BERC. De forma geral, quanto maior BERC, menor ? a chance de o sinal ser decodificado por um receptor n?o autorizado. Na decodifica??o o sinal ? novamente dividido em fatias espectrais, e para todas as fatias s?o aplicados valores distintos de atenua??o e atraso de forma a reconstituir o sinal de entrada. Na sa?da do decodificador, avalia-se a BER do sinal decodificado BERD, que deve ser suficientemente baixa para permitir a recep??o do sinal transmitido pela TON. Simula??es com o software VPITransmissionMaker, da empresa VPIPhotonics Inc, foram utilizadas para investigar o desempenho da t?cnica para diferentes ganhos e atrasos. Para a opera??o de fatiamento espectral, os perfis dos filtros usados foram ajustados para representar equipamentos de mercado. Resultados indicam que BERC pode atingir valores de at? 42% e 32% para sinais com modula??o non return to zero - on-off keying (NRZ-OOK) e por deslocamento de fase diferencial em quadratura (differential quadrature phase shift keying, DQPSK), respectivamente. Em ambos os casos foi poss?vel encontrar resultados de BERd inferiores a 10-12, mediante ajuste adequado do espa?amento entre os filtros, mesmo ap?s 400 km de propaga??o por enlaces de fibra padr?o. No melhor de nosso conhecimento, esta ? a primeira vez que tais an?lises s?o apresentadas.