RESUMEN
We experimentally explore the practicality of integrated multiwavelength laser arrays (MLAs) for photonic convolutional neural network (PCNN). MLAs represent excellent performance for PCNN, except for imperfect wavelength spacings due to fabrication variation. Therefore, the performance of PCNN with non-ideal wavelength spacing is investigated experimentally and numerically for the first time. The results show that there exists a certain tolerance for wavelength deviation on the degradation of the structural information of the extracted feature map, leading to the robustness of photonic recognition accuracy under non-ideal wavelength spacing. The results suggest that scalable MLAs could serve as an alternative source for the PCNN, to support low-cost optical computing scenarios. For a benchmark classification task of MNIST handwritten digits, the photonic prediction accuracy of 91.2% for stride 1 × 1 scheme using the testing dataset are experimentally obtained at speeds on the order of tera operations per second, compared to 94.14% on computer. The robust performance, flexible spectral control, low cost, large bandwidth and parallel processing capability of the PCNN driven by scalable MLAs may broaden the application possibilities of photonic neural networks in next generation data computing applications.
RESUMEN
We describe the mathematical model and present simulation results for the optimization of a hybrid Raman/optical parametric amplifier (HROPA), exhibiting a bandwidth of 170 nm and low ripple that covers the top half of the wavelength plan (e.g., 1441 to 1611 nm) of next generation coarse wavelength division multiplexed passive optical network systems. We show that a critical parameter in the proper amplifier parameter optimization is the inclusion of the fourth-order dispersion coefficient (ß(4)). Omission of ß(4) can lead to over-estimation or underestimation of the gain bandwidth, and hence its inclusion in the analysis of the HROPA is necessary.
RESUMEN
At a reflective intensity modulated fiber optic microphone (RIM-FOM), the acoustic signal makes a membrane vibrate and modulate the reflected intensity. In the existing models of the RIM-FOM, the offset of all points of the membrane, due to the vibration, is assumed to be equal. However, this assumption does not represent the actual vibration of the membrane, which follows a continuous surface shape change caused by the acoustic signal. We establish a revised theoretical model in which the influence of the actual membrane surface shape change on the reflective intensity modulation is considered. Experiments show that there is a discrepancy between the experimental optimum operating distance and the analytical result from the existing model, while our new model gives a better agreement with the experimental results. In particular, our analysis shows that, in using the existing model, the other microphone performance characteristics are misestimated, while our revised model can provide a closer solution.
RESUMEN
A long-period fiber grating (LPFG) humidity sensor is reported utilizing poly(ethylene oxide)/cobalt chloride (PEO/CoCl2) as a hybrid hygrosensitive cladding coating. A thin overlay of the material is deposited on the LPFG and with exposure to different ambient humidity levels, its spectral properties are modified. The material parameters associated with the sensing mechanism may include those of refractive index, absorption, and morphological alterations of the overlaid material. Relative humidity variations in the range from 50% to 95% have been detected with a resolution better than 0.2%. The response time constant of the fiber sensor is of the order of a few hundred milliseconds.