The packet scheduling in router plays an important role in the sense to achieve QoS differentiation and to optimize the queuing delay, in particular when this optimization is accomplished on all routers of a path between source and destination. In a dynamically changing environment a good scheduling discipline should be also adaptive to the new traffic conditions. We model this problem as a multi-agent system in which each agent learns through continual interaction with the environment in order to optimize its own behaviour. So, we adopt the framework of Markov decision processes applied to multi-agent system and present a pheromone-Q learning approach which combines the Q-multi-learning technique with a synthetic pheromone that acts as a communication medium speeding up the learning process of cooperating agents. 相似文献
Stochastic gradient descent (SGD) is a widely adopted iterative method for optimizing differentiable objective functions. In this paper, we propose and discuss a novel approach to scale up SGD in applications involving non-convex functions and large datasets. We address the bottleneck problem arising when using both shared and distributed memory. Typically, the former is bounded by limited computation resources and bandwidth whereas the latter suffers from communication overheads. We propose a unified distributed and parallel implementation of SGD (named DPSGD) that relies on both asynchronous distribution and lock-free parallelism. By combining two strategies into a unified framework, DPSGD is able to strike a better trade-off between local computation and communication. The convergence properties of DPSGD are studied for non-convex problems such as those arising in statistical modelling and machine learning. Our theoretical analysis shows that DPSGD leads to speed-up with respect to the number of cores and number of workers while guaranteeing an asymptotic convergence rate of \(O(1/\sqrt{T})\) given that the number of cores is bounded by \(T^{1/4}\) and the number of workers is bounded by \(T^{1/2}\) where T is the number of iterations. The potential gains that can be achieved by DPSGD are demonstrated empirically on a stochastic variational inference problem (Latent Dirichlet Allocation) and on a deep reinforcement learning (DRL) problem (advantage actor critic - A2C) resulting in two algorithms: DPSVI and HSA2C. Empirical results validate our theoretical findings. Comparative studies are conducted to show the performance of the proposed DPSGD against the state-of-the-art DRL algorithms.
The development of eco-friendly connection material instead of steel is a challenging problem in timber structures. Following densification, the mechanical properties of low-density species can be significantly improved. Densified wood may be a potential connection material in timber structures. This paper reviewed the different processing for densified wood, and obtained favorable mechanical properties and dimensional stability based on small specimen sizes, which are much less than the applicable sizes in practice. A densification processing with alkali pretreatment was adopted for poplar widely cultivated in the world to produce the densified poplar, which has been rarely reported as connection material. Various specimens of densified poplar were tested to obtain their main mechanical properties such as strength and deformability. The set recovery of densified poplar was also measured to observe their dimensional stability. In addition, the hygroscopic swelling strains for the diameter of densified poplar dowel were measured to present their moisture-dependent behavior. The improved mechanical properties and dimensional stability confirmed the fact that densified poplar with alkali pretreatment can be an optimal connection material.
Privacy in Vehicular Ad Hoc Networks (VANET) is fundamental because the user's safety may be threatened by the identity and the real‐time spatiotemporal data exchanged on the network. This issue is commonly addressed by the use of certified temporal pseudonyms and their updating strategies to ensure the user's unlinkability and anonymity. IEEE 1609.2 Standard specified the process of certifying pseudonym along with certificates structure. However, the communication procedure between the certifying authority and the requesting vehicle was not defined. In this paper, a new privacy‐preserving solution for pseudonym on‐road on‐demand refilling is proposed where the vehicle anonymously authenticates itself to the regional authority subsidiary of the central trusted authority to request a new pseudonyms pool. The authentication method has two phases, the first one uses anonymous tickets, and the second one is a challenge‐based authentication. The anonymous tickets are certificates that do not include the identity of the user. Instead, it contains a reference number and the certifying authority signature. The challenge authentication is identity‐less to preserve the privacy, yet it is used to prevent the misuse of tickets and the impersonation of its owner. Our proposed scheme is analyzed by the use of Burrows, Abadi and Needham (BAN) logic to demonstrate its correctness. It is also specified and checked by using the Security Protocol ANimator (SPAN) and the Automated Validation of Internet Security Protocols and Applications (AVISPA) tools. The logical demonstration proved that this privacy‐preserving authentication is assured. The SPAN and AVISPA tools illustrated that it is resilient to security attacks. 相似文献
The use of vegetable fibers has regained interest in different fields. In fact, fibers may represent an alternative to traditional fibers incorporated in building materials, such as polypropylene fibers. Sustainable development requires the adoption of ecofriendly and natural materials. In this approach, using vegetable fibers as reinforcement for cementitious materials represents an interesting option for concrete industry. Thus, this article studies mechanical performances and transfer properties of cementitious mortars reinforced with vegetable fibers. The used fibers were Dis (DS), Alfa (AF), Date palm (DP) and Hemp (HE). They were subjected to prior treatment with calcium hydroxide. The results show that reinforced mortars with vegetable fibers represent a better deformation capacity than the control mortar without fibers; the best bending strength was recorded for hemp and date palm mortars; the addition of vegetable fibers leads to thermal conductivity decrease by 15% for a content of 0.1%. However, greater carbonation depths were recorded for vegetable fibers based mixes. These findings, for us, are promising for the use of vegetable fibers in cementitious materials. 相似文献
Thin films of silicon nitride (Si3N4) can be used in several kinds of micro-sized biosensors as a material to monitor fine environmental changes related to the
process of bone formation in vitro. We found however that Si3N4 does not provide optimal conditions for osseointegration as osteoblast-like MG-63 cells tend to detach from the surface when
cultured over confluence. Therefore Si3N4 was modified with self-assembled monolayers bearing functional end groups of primary amine (NH2) and carboxyl (COOH) respectively. Both these modifications enhanced the interaction with confluent cell layers and thus
improve osseointegration over Si3N4. Furthermore it was observed that the NH2 functionality increased the adsorption of fibronectin (FN), promoted cell proliferation, but delayed the differentiation.
We also studied the fate of pre-adsorbed and secreted FN from cells to learn more about the impact of above functionalities
for the development of provisional extracellular matrix on materials interface. Taken together our data supports that Si3N4 has low tissue integration but good cellular biocompatibility and thus is appropriate in cellular biosensor applications
such as the ion-sensitive field effect transistor (ISFET). COOH and NH2 chemistries generally improve the interfacial tissue interaction with the sensor and they are therefore suitable substrates
for monitoring cellular growth or matrix deposition using electrical impedance spectroscopy. 相似文献
Optical absorption in thin-film solar cells can be improved by using surface plasmons for guiding and confining the light on the nanoscale. We report theoretical and simulation studies of a-Si thin-film solar cells with silver nanocylinders on the surface. We found that surface plasmons increased the cells'' spectral response over almost the entire studied solar spectrum. In the ultraviolet range and at wavelengths close to the Si band gap we observed a significant enhancement of the absorption for both thin-film and wafer-based structures. We also performed optimization studies of particle size, inter-particle distance, and dielectric environment, for obtaining maximal absorption within the substrate. A blue-shift of the resonance wavelength with increasing inter-particle distance was observed in the visible range. Cell performance improved at optimal spacing, which strongly depended on the nanoparticle size. Increasing the nanocylinder size was accompanied by the widening of the plasmon resonance band and a red-shift of the plasmon resonance peaks. A weak red-shift and plasmon peak enhancement were observed in the reflectance curve with increasing refractive index of the dielectric spacer. 相似文献