Search

With the advent of IoT, as an extension of the traditional cloud-centric computing model, the edge computing model solves the network channel congestion caused by the increasing number of IoT devices and satisfies theirs low latency requirements. Since the edge network needs to adopt heterogeneous structure to meet multiple access requirements during construction, how to properly allocate limited edge resources to process computing requests from mobile edge terminals becomes a challenge under heterogeneous networks. In recent years, reinforcement learning shine in various types of decision-making problems, it has brought new ideas to solve complex decision problems. In this paper, we use reinforcement learning to solve the resource allocation problem of heterogeneous edge server (ES Edge Server) collaborative computing. In this model, we consider the multitasking and the heterogeneity and mobility of edge devices, abstracting the problem into Markov Decision Processes (MDPs) based on the real-time state of the network and the attributes of the task, using based on Actor Critic and Policy Gradient’s DDPG (Deep Deterministic Policy Gradient) make resource allocation optimization decisions. The simulation results show that compared with the DQN-based algorithm, the DDPG-based algorithm has better simulation results in the experiment.

Bandwidth demand prediction is not only the basis for the preparation of communication network planning, but also an important prerequisite for optimizing the efficiency of resource allocation and maximizing the efficiency of communications support. This paper explores the problems related to the prediction of transmission bandwidth demand of private network communication hub stations. The author comprehensively analyzes the factors influencing the measurement of transmission bandwidth demand by adopting a service-varying prediction summation method method based on QOS guarantee. By studying the business type models, the author builds a model for calculating the bandwidth demand in access layerof the transmission hub, and selects typical stations in the private network to verify the calculation model. The simulation result shows that when the service quality is ensured, the proposed prediction method can be used to predict the transmission bandwidth in an accurate manner, thereby providing a reference for the optimization of private network planning and resource allocation.

At present, the demand response resources of large users are gradually full, and the existing ternary market model of thermal storage electric heating is difficult to manage scattered small and medium-sized loads and guide users to participate in the power market. In this study, by introducing a load aggregator, the original ternary market model was transformed into a quaternary market transaction model of new energy enterprises — grids — load aggregators — regenerative electric heating users. Based on the transaction cost theory, the analysis introduced load aggregators The mechanism of the new post-business model on the electricity trading market. It proved that the new model can help reduce transaction costs and promote electricity market transactions.

Ganoderma fungi have long been used as functional foods and traditional medicines in Asian countries. Ganoderma ahmadii is one of the main species of Ganoderma fungi distributed in Hainan province of China, the fruiting bodies of which have been used in folk to lower blood sugar for a long time. A chemical investigation of the fruiting bodies of Ganoderma ahmadii led to the isolation of seven new meroterpenoids, named ganoduriporols F-L (1–7). The chemical structures of the compounds were elucidated by spectroscopic data including HRESIMS and 2D NMR. Compounds 5–7 represent the first examples of ganoduriporol-type meroterpenoids bearing oxepane rings in their skeletons. Compounds 1–4 showed inhibitory activity against protein tyrosine phosphatase 1B (PTP1B) comparable to the positive control Na3VO4, with IC50 values of 17, 20, 19, and 23 μM, respectively.

Recently, convolutional neural network (CNN) based methods have achieved significantly better performance compared to conventional methods based on hand-crafted features for image steganalysis. However, as far as we know, existing CNN based methods extract features either with constrained (even fixed), or random (i.e., randomly initialized) convolutional kernels, and this leads to limitations as follows. First, it is unlikely to obtain optimal results for exclusive use of constrained kernels due to the constraints. Second, it becomes difficult to get optimal when using merely random kernels because of the large parameter space to learn. In this paper, to overcome these limitations, we propose a two-way convolutional neural network (TCNN) for image steganalysis, by combining both constrained and random convolutional kernels, and designing respective sub-networks. Intuitively, by complementing one another, the combination of these two kinds of kernels can enrich features extracted, ease network convergence, and thus provide better results. Experimental results show that the proposed TCNN steganalyzer is superior to the state-of-the-art CNN-based and hand-crafted features-based methods, at different payloads.

International audience; Confined glasses and their anomalous interfacial rheology raise important questions in fundamental research and numerous practical applications. In this Letter, we study the influence of interfacial air nanobubbles on the free surface of ultrathin high-molecular-weight glassy polystyrene films immersed in water, in ambient conditions. In particular, we reveal the counterintuitive fact that a soft nanobubble is able to deform the surface of a rigid glass, forming a nanocrater with a depth that increases with time. By combining in-situ atomic-force-microscopy measurements and a modified lubrication model for the liquid-like layer at the free surface of the glass, we demonstrate that the capillary pressure in the nanobubble together with the liquid-like layer at the free surface of the glass determine the spatiotemporal growth of the nanocraters. Finally, from the excellent agreement between the experimental profiles and the numerical solutions of the governing glassy thin-film equation, we are able to precisely extract the surface mobility of the glass. In addition to revealing and quantifying how surface nanobubbles deform immersed glasses, until the latter eventually dewet from their substrates, our work provides a novel, precise, and simple measurement of the surface nanorheology of glasses.

A wear-resistant Fe-Mo alloy coating was fabricated on the AISI 1045 steel substrate by plasma transferred arc (PTA) cladding, using pure Mo powders as the precursor material. The microstructure, microhardness and wear resistance of the coating were investigated by scanning electron microscope (SEM) with energy dispersive spectrum (EDS), X-ray diffraction (XRD), Vickers hardness and pin-on-disk wear testers, respectively. It was found that the coating presented rapidly solidified microstructure mainly consisting of cellular R-Fe63Mo37 dendrites and a small amount of lamellar R-Fe63Mo37/α-Fess dendrites uniformly distributed in the α-Fess matrix. Based on the eutectic growth and undercooling theories, the possibility of the R/α peritectic coupled growth in the rapidly solidified peritectic Fe-Mo alloy was discussed. The extensive growth of columnar crystallites into the substrate provided good metallurgical bonding at the interface between the coating and the AISI 1045 steel substrate. The results showed that the microhardness and wear resistance of the coating were much higher than that of the AISI 1045 steel substrate. The transferred oxides layer containing MoO3 formed on the worn surface had a good friction-reducing property and also contributed to the wear resistance of the coating.

The evolution behaviors and mechanisms of internal crack healing in three different kinds of steels, including 20 high-quality carbon structural steel, 30Cr2Ni4MoV steel, and SUS304 stainless steel, at elevated temperatures were systematically investigated. The morphology of the crack healing zone was observed using an optical microscope (OM) and a scanning electron microscope (SEM). The grain boundary character distribution evolution of the crack healing zone was investigated using electron backscatter diffraction (EBSD). The results show that the existence of two crack healing mechanisms is confirmed in all three experimental steels. Crack healing is controlled by atomic diffusion at lower temperatures 1173 K (≤ 900 °C) and depends mainly on recrystallization and grain growth mechanisms at higher temperatures 1273 K (≥ 1000 °C). Atomic diffusion provides materials for recrystallization and grain growth in the crack healing zone. Recrystallization leads to rapid crack healing, and grain growth facilitates microstructural homogenization and elimination of the crack healing zone. The process of internal crack healing can be divided into five stages.

The microstructure evolution of high nitrogen austenitic steel wires under various annealing times and drawing temperatures was carefully characterized. Special attention was paid to the widely distributed twins and the nanoprecipitates at twin boundaries (TBs) in high nitrogen stainless steels (HNSSs). The results of microhardness indicated that the traditional Hall–Petch (H–P) equation, which only took the role of grain boundaries into account, was unsuitable. A new H–P equation that connected grain size, twin density, precipitates at TBs, and microhardness in HNSS was established for the first time and showed to be in good agreement with the experimental results. By analyzing the strained regions near TBs, a model describing the precipitation of nano-M23C6 carbides on coherent twin boundaries and incoherent twin boundaries was proposed. In addition, the influence mechanism of the nano-M23C6 at TBs on microhardness was discussed.