基于平均时间的线程池尺寸自适应调整算法

为了实现自适应调整线程池尺寸,提高并发程序处理的运行速度、改善效率和降低系统的资源开销,提出了一种基于任务平均处理时间的线程池尺寸自适应调整算法。首先研究线程池的一些特征数据以及用户请求的任务时间周期,提出了任务平均处理时间的概念。然后研究了不同任务类型下,任务平均处理时间与线程池的尺寸的相关性,提出了一种自适应调整算法。实验结果表明,该算法能够自适应调整线程池尺寸到适当区域,有效地提高应用程序的整体性能。     

基于分段的线程池尺寸自适应调整算法

提出一种基于分段的线程池尺寸自适应调整算法。该算法将用户请求量分为上升段、平衡段和下降段3段,根据当前用户请求数、线程数自适应调整线程池尺寸,从而满足用户需求。实验结果表明,相比基于平均数的调整算法,该算法能更好地处理并发的用户请求,响应时间更短。     

线程池技术应用研究

目前几乎所有操作系统都支持多线程的工作方式,以提高系统的处理能力.本文研究了采用线程池技术可以提高服务器程序性能,并通过一个实例分析了线程池技术中线程池尺寸和任务数对服务器程序的性能影响.     

一种有效的储备池在线稀疏学习算法

为克服传统储备池方法缺乏良好在线学习算法的问题, 同时考虑到储备池本身存在的不适定问题, 本文提出一种储备池在线稀疏学习算法, 对储备池目标函数施加L1正则化约束,并采用截断梯度算法在线近似求解.所提算法在对储备池输出权值进行在线调整的同时, 可对储备池输出权值的稀疏性进行有效控制, 有效保证了网络的泛化性能.理论分析和仿真实例证明所提算法的有效性.     

基于多重指数移动平均评估的DDPG算法

针对深度确定性策略梯度算法中双网络结构的不稳定及单评论家评估不准确的问题,提出基于多重指数移动平均评估的DDPG算法.介绍EM A-Q网络和目标Q网络合作得出目标更新值,对多个评论家给出的Q值求平均,降低单评论家评估的不准确性.样本池部分引入双重经验回放方法,采用两个样本池分别存储不同的经验,提高算法的收敛性能.将所提算法及原始DDPG算法分别实验于经典的Pendulum问题和Mountain Car问题中.实验结果表明,与传统的DDPG算法相比,所提算法准确性更好,稳定性更高,收敛速度明显提升.     

多元时间序列的子空间回声状态网络预测模型

针对采用回声状态网络预测多元混沌时间序列时储备池学习算法可能存在的病态解问题,该文提出了一种基于快速子空间分解方法的回声状态网络预测模型.所提模型利用Krylov子空间分解方法提取储备池状态矩阵的子空间,子空间代替原状态矩阵进行输出权值求解,可以消除储备池状态矩阵的冗余信息,有效地解决伪逆算法存在的病态解问题,并且降低计算复杂度,提高泛化性能和预测精度.基于两组多元混沌时间序列的仿真结果验证了该文所提模型的有效性和实用性.     

用混合遗传算法求解集团现金池收益优化问题

集团现金池业务的收益优化决策一直是由人工辅助完成,存在决策效率低下、全局寻优能力差等问题.首次将现金池财务成本优化模型抽象为一个资源受限项目调度问题RCPSP(Resource Constrained Project Scheduling Problem),并引入结合了遗传算法全局寻优能力以及模拟退火算法局部寻优能力的混合算法解决了此NP难问题.仿真示例证明了该现金池优化模型和混合遗传算法的有效性.     

基于改进型SSD的视频烟火检测算法

针对现有烟火检测算法对小尺寸烟火目标检测效果差,容易产生漏检和误检的问题,提出一种基于改进型SSD的视频烟火检测算法.利用DenseNet网络作为SSD的基础网络,提高其对小目标的检测能力.为了改进SSD中的正负样本不平衡的问题,在损失函数中引入Focal loss函数,通过提高难分样本在损失函数中的权重,提高算法的鲁棒性.在仿真实验中,通过构建烟火数据集对改进的SSD模型进行训练.实验结果表明,相比现阶段主流的几种目标检测算法,该算法在兼顾检测速度的同时提高了对小尺寸烟火目标的检测效果.     

一种非局部连接的抗噪性储备池构建方法

由于具有高度的生物真实性, 液体状态机在抗噪性、鲁棒性方面相对于人工神经网络具有更大的优势, 但 也更难优化. 采用人工神经网络思想对液体状态机进行的优化, 牺牲了生物真实性和网络泛用性的同时, 并不能保 证优化的有效性; 而依据生物神经系统内抽象出的规律进行储备池的优化, 则优化算法非常复杂. 为了提高储备池 的泛用性和抗噪性, 同时避免复杂的优化过程, 本文模拟大脑中普遍存在的各神经元集群间的非局部连接分布—伽 马分布来生成储备池的权值, 生成一个具有更高生物真实性、隐含功能柱结构的储备池. 首先, 通过对储备池活动 和储备池进行Lempel-Ziv复杂度分析, 从理论上说明该种储备池权值生成方式的优势; 然后, 通过与脉冲时序可塑 性算法(STDP)和高斯分布等进行对比实验, 证明本文采用伽马分布生成的储备池具有更高的准确度和更强的抗噪 性.     

APHMIPv6:快速移动IPv6切换算法

当移动节点发生切换时,其切换时延和丢包率问题将严重影响实时业务的应用.针对这一问题,该文进行了理论分析并提出了一种基于地址池的层次型移动IPv6切换算法(APHMIPv6).该算法与层次型模型相结合,采用基于地址池的有状态地址配置策略以及域间切换时利用隧道机制来提高切换性能.仿真结果表明,该算法能有效地减少移动节点切换时延和丢包率,尤其在域间切换时,性能更加显著.     

A mechanism for resource pricing and fairness in peer-to-peer networks

In peer-to-peer (P2P) networks, each peer acts as the role of client and server. As a client, each peer is regarded as a service customer. It sends requests to other peers to download files and obtains resource allocation from them. As a server, each peer is thought as a service provider. It receives service requests from other peers and allocates its resources to them. To encourage cooperation between peers, fairness is very important in P2P networks since it fosters an incentive to the peers to offer resources to the network. We formulate a fair resource allocation model for P2P networks and investigate the utility optimization problem by Lagrangian method. In order to realize the optimal resource allocation, we present a novel price-based resource allocation scheme by applying the first order Lagrangian method and low-pass filtering scheme, so that a service provider can allocate its resources to its customers based on offered prices, achieving the efficient and fair allocation of the available resources to the serviced customers. Simulation results confirm that the proposed algorithm can achieve the optimum within reasonable convergence times.     

Multi-Layer Resource Management in Cloud Computing

The paper studies multi-layer optimization in service oriented cloud computing to optimize the utility function of cloud computing, subject to resource constraints of an IaaS provider at the resource layer, service provisioning constraints of a SaaS provider at the service layer, and user QoS (quality of service) constraints of cloud users at application layer, respectively. The multi-layer optimization problem can be decomposed into three subproblems: cloud computing resource allocation problem, SaaS service provisioning problem, and user QoS maximization problem. The proposed algorithm decomposes the global optimization problem of cloud computing into three sub-problems via an iterative algorithm. The experiments are conducted to test the efficiency of the proposed algorithm with varying environmental parameters. The experiments also compare the performance of the proposed approach with other related work.     

A new scheme for traffic estimation and resource allocation for bandwidth brokers

This paper is motivated by the concern of a multi-service network provider who plans to offer quality of service guarantees to users. A bandwidth broker acts as the resource manager for each network provider. Neighboring bandwidth brokers communicate with each other to establish inter-domain resource reservation agreements. Conventional approaches for resource allocation rely on pre-determined traffic characteristics. If allocation follows the traffic demand very tightly, the resource usage is efficient but leads to frequent modifications of the reservations. This would lead to increased inter-bandwidth-broker signaling in order to propagate the changes to all the concerned networks. Contrarily, if large cushions are allowed in the reservations, the modifications are far spaced in time but the resource usage becomes highly inefficient. In this paper, a new scheme for estimating the traffic on an inter-domain link and forecasting its capacity requirement, based on a measurement of the current usage, is proposed. The method allows an efficient resource utilization while keeping the number of reservation modifications to low values.     

采用随机规划模型的云资源分配算法

针对云资源提供问题,为了降低云消费者的资源使用成本,提出了一种采用随机规划模型的云资源分配算法.同时考虑按需实例和预留实例,采用两阶段随机整数规划对云资源提供问题进行建模,在资源预留阶段,根据长期的工作负载情况,确定预留实例的类型和数量,在按需分配阶段,根据当前的工作负载,确定动态分配的按需实例的类型和数量.采用抽样平均近似方法减少资源提供问题的场景数量,降低求解复杂度,并提出了一种基于阶段分解的混合进化算法求解资源提供问题.仿真实验结果表明,采用随机规划模型的云资源分配算法能够在较短时间内获得近似最优的云资源预留方案,有效降低了云消费者的资源使用成本.     

Dynamic service-level-agreement aware shared-path protection in WDM mesh networks

The notion of service level agreements (SLA) has been proposed to capture qualitatively and quantitatively defined performance contracts between service providers and their customers. How to provision satisfiable connections for customers according to their SLA parameters to avoid penalty as well as minimizing cost is one of the main concerns for a service provider. Connection reliability and restoration time are two important SLA parameters for a customer, and should be carefully considered in survivable wavelength-division-multiplexing (WDM) networks. A sound scheme should guarantee customers’ reliability and reduce restoration time while benefiting a service provider in resource efficiency. Under the SLA constraints and the assumption of single link failure, we first investigate a partial link-disjoint protection (PLDP) and analyze its availability and average restoration time. We then propose a dynamic constraint shared path protection (DCSP) algorithm in WDM mesh networks. Based on the basic ideas of the K-shortest path algorithm and PLDP, DCSP can provide differentiated services for customers according to their SLA parameters while being favorable for reducing average restoration time and optimizing resource utilization. Simulation results show that DCSP can efficiently guarantee the specific SLA requirements of customers with a slight increase in the blocking probability while still achieving spare resource sharing and restoration time optimizations.     

Logistics service provider selection decision making for healthcare industry based on a novel weighted density-based hierarchical clustering

In recent years, Industry 4.0 makes a significant impact on the manufacturing industry, which enables the business more intelligent and efficient, all while minimizing costs. As known, the logistics concerns in the supply chain always play an important role to a manufacturing company, and decision on the selection of logistics service provider is a key point, especially for healthcare manufacture whose products are the medical devices or equipment of fragility and high cost. Practically there are so many logistics service providers with varieties in service quality, effectiveness, punctuality and reliability, that the manufacturers often encounter the challenge on the provider selection, and healthcare industry is no exception. However, the research on provider selection for healthcare manufacturers is quite limited. In order to help them to make the decision, this paper designs a logistics service provider selection scheme based on a novel weighted density-based hierarchical cluster analysis with integration of the analytic hierarchy process (AHP) for healthcare industry. Initially an evaluation index system reflecting the capability of the candidate providers in all aspects is established. To improve the clustering within the scheme, the density concept and the obtained weights are introduced into the traditional hierarchical cluster analysis (HCA) to shape a novel Weighted Density-Based HCA (WDBHCA). To validate the feasibility of the scheme, a case study on a specified healthcare industry manufacturer is carried out, and results fulfill the case company’s requirement which shows the feasibility of the proposed provider selection scheme. In addition, this scheme can be applied to the provider selection in other fields, as well.     

A distributed architecture for policy-customisable multi-tenant processes-as-a-service

Service-based business processes are often developed and deployed by single organisations. In distributed, shared resource environments like the cloud on the other hand, consumers share resources owned by cloud providers. This requires multi-tenancy capability for service processes that provide customised behaviour on shared process implementations to meet the varying needs of different process consumers as tenants of the process resource. In this paper, we define a distributed multi-tenant architecture for Business Process Execution Language (BPEL) processes provided as a service. A single-version BPEL process is deployed by a provider and offered for all process consumers, combined with a customisation and management functionality to create a unique experience for different consumers (process tenants). We provide two core components: a policy model for consumers to express customisation/business requirements of service processes and a coordination framework for policy enforcement between consumers and providers to achieve on-the-fly customisation of service processes.     

云计算下基于CRP算法的资源提供策略

云计算资源管理系统是用于接收来自云计算用户的资源请求,并且把特定的资源封装为服务提供给资源请求者。在云计算环境下,如何为资源请求者选择合适的资源是一个值得研究的课题。文中通过对云计算下现有的资源提供策略的分析,同时根据不同云提供者提供的计算资源的成本不同的特点,综合考虑资源的计算能力、可靠性和单位成本三点因素,提出了云计算下基于CRP算法的资源提供策略。这种资源提供策略既能提供满足用户资源请求的服务,也能降低云服务提供者的运营成本,从而获得更大收益。     

面向边缘计算的组合拍卖式任务卸载机制

在万物互联的时代,数据量与计算需求飞速增长,促使应用部署方式由云计算模式向边缘计算模式演进,以解决带宽消耗严重和响应时延过高等问题。为推进面向边缘网络的任务卸载,需要解决应用服务提供商(ASP)与边缘计算提供商(ECP)之间的双向选择问题。针对这一问题,提出一种面向边缘计算的组合拍卖式任务卸载机制。首先建立系统模型,并对模型落地的关键问题进行说明,然后分析ECP的投标决策过程,证明选择最大化资源利用率的任务组合是NP完全问题,进而提出一种启发式任务选择算法。在此基础上,设计两种拍卖算法,单胜者拍卖和多胜者拍卖,分别适用于可信度优先和效率优先的场景。实验结果表明,相较于单项拍卖机制,所提出的方案提高ECP资源利用率达13%,同时增加ASP收益达37%。