Elastic optical network (EON) is a promising technology for the next generation of optical transport network. Routing, modulation and spectrum assignment is one of the key issues in EON. In the problem, the constraints of spectrum continuity and contiguity must be satisfied. In this paper, we propose a spectrum assignment policy named minimizing sum of weighted resource reductions (min $\sigma w \Delta R$ ) algorithm. The algorithm includes three contributions: 1) The concept of resource reduction ( $\Delta R$ ) is defined to measure variation of idle resources in a path when a spectrum block is assigned to a connection request, which is combination of changes of continuity and contiguity in the path. 2) When a spectrum block in a path is used, resource reductions in all associated paths rather than only the current path are taken into account. 3) A function of path's hop is used as a weight to adjust the path's impact. Simulations were carried out based on the national science foundation network (NSF-Net) and the pan-European Cost239 network (Cost239) to investigate the performance of the algorithm, by comparing with three well-performed benchmarks. Simulation results show that, when the parameters of the metric are set to appropriate values, the proposed algorithm can significantly decrease blocking probability in both two topologies.

A magnetically recoverable biocatalyst has been effectively prepared through the immobilization of α-amylase onto the Fe₃O₄ nanoparticles. The magnetic nanoparticles (MNPs) were synthesized by a sol–gel method in an aqueous system. The MNPs’ surfaces were modified with sodium silicate and 3-aminopropyltriethoxysilane (APTS). The adsorptive immobilization of α-amylase was attempted onto the APTS–SiO₂–Fe₃O₄ nanoparticles. The physicochemical properties of the biocatalysts were characterized with Fourier transform infrared spectroscopy (FTIR), transmission electron microscopy (TEM), X-ray diffraction (XRD) and thermal gravimetric analysis (TGA). The magnetic properties of the biocatalyst were analyzed by a vibrating sample magnetometer (VSM). The activity of free and immobilized α-amylase was also evaluated by direct measurement of the absorbance of starch solution at 620 nm using UV-vis spectrophotometry. The prepared particles exhibited superparamagnetic behavior at room temperature. The mass saturation magnetization (M_s) of the bare Fe₃O₄ nanoparticles and the biocatalyst was 175.7 and 50.4 emu g⁻¹ respectively. A biocatalyst was produced with a loading capacity of α-amylase of about 235 mg g⁻¹. The enzymatic activity of the immobilized enzyme was 882 U g⁻¹ which is about 79.53% of the free enzyme activity when not involved in the immobilized system. The immobilized enzyme exhibited significant thermal stability (stable up to 70 °C) and good durability (recycled 3 times without any obvious loss of enzymatic activity). This synthetic method provides a possible industrial route for the use of enzyme catalysts which eliminate the high operational costs and complicated recycle/reuse problems in the conventional enzyme catalysis process.

Deep neural networks are extremely vulnerable to attacks and threats from adversarial examples. These adversarial examples deliberately crafted by attackers can easily fool classification models by adding imperceptibly tiny perturbations on clean images. This brings a great challenge to image security for deep learning. Therefore, studying and designing attack algorithms for generating adversarial examples is essential for building robust models. Moreover, adversarial examples are transferable in that they can mislead multiple different classifiers across models. This makes black-box attacks feasible for practical applications. However, most attack methods have low success rates and weak transferability against black-box models. This is because they often overfit the model during the production of adversarial examples. To address this issue, we propose a Nadam iterative fast gradient method (NAI-FGM), which combines an improved Nadam optimizer with gradient-based iterative attacks. Specifically, we introduce the look-ahead momentum vector and the adaptive learning rate component based on the Momentum Iterative Fast Gradient Sign Method (MI-FGSM). The look-ahead momentum vector is dedicated to making the loss function converge faster and get rid of the poor local maximum. Additionally, the adaptive learning rate component is used to help the adversarial example to converge to a better extreme point by obtaining adaptive update directions according to the current parameters. Furthermore, we also carry out different input transformations to further enhance the attack performance before using NAI-FGM for attack. Finally, we consider attacking the ensemble model. Extensive experiments show that the NAI-FGM has stronger transferability and black-box attack capability than advanced momentum-based iterative attacks. In particular, when using the adversarial examples produced by way of ensemble attack to test the adversarially trained models, the NAI-FGM improves the success rate by 8% to 11% over the other attack methods. Last but not least, the NAI-DI-TI-SI-FGM combined with the input transformation achieves a success rate of 91.3% on average.

Elastic optical networks (EONs), which can flexibly assign spectrum resources to connection requests, have become the most promising technology for the next generation of optical core networks. Routing, modulation and spectrum assignment (RMSA) problem is one of the key issues in elastic optical networks. In the RMSA problem, spectrum resources in links are generally divided into frequency slices and connection requests with different capacities can use different numbers of frequency slices to transfer their data; however, this may lead to misalignment of idle frequency slices in links and then result in the increase of blocking probability of connection requests. In this paper, we propose a resource-periodic-arrangement (RPA) strategy. The strategy partitions frequency slices into periodically arranged standard frequency blocks and each connection request exactly uses one or more standard frequency blocks. The RPA strategy can effectively reduce the misalignment of frequency slices. Furthermore, an RPA-based RMSA algorithm is designed, in which an enhanced-most-used policy is adopted to choose standard frequency blocks to connection requests. Simulation results show that the RPA-based RMSA algorithm can distinctly reduce blocking probability of connection requests, comparing with existing algorithms.