Optical packet header identification utilizing an all-optical feedback chaotic reservoir computing.

In this paper, an all-optical reservoir computing (RC) setup is proposed for identifying the types of optical packet headers in optical packet switching (OPS) network. The numerical simulation identification results of 3 bits and 32 bits optical headers with the bit rate of 10 Gbps are as low as 0.625% and 2.25%, respectively. The identification errors with the variation of the feedback strength and feedback delay are presented separately. Hence, the optimal feedback parameters are obtained. The all-optical feedback RC setup is robust to the white Gaussian noise. The recognition error is acceptable when the signal-to-noise ratio (SNR) is greater than 15 dB. [ABSTRACT FROM AUTHOR]