Next Generation Real-Time Methods for the Supervision and Contol of Industrial Environments

这一研究应用的目的是调查与设计和实施用于监督控制企业流程的下一代信息技术基础设施的具体组件有关的选定的理论和实验主题。众所周知，由于计算机和网络技术的商品化，信息和通信技术是新的数字经济的核心部分。最近，世界经济论坛预测，新数字经济将成为任何发达国家的主要产业组成部分。越来越多的应用程序和数据正在从客户驻地迁移到云环境。报告显示，平均一个云计算基础设施消耗的能源相当于25,000个家庭。过去几年在分布式和云计算领域取得的进展导致了由成本极高的大型数据中心托管的大规模软件技术的产生，而“绿色计算”运动促进了提高云的能源消耗效率。尽管进行了持续的研究工作，但由于计算、存储和网络资源的利用，云计算基础设施中的能源消耗动态的正式模型仍然缺乏。我们提出用随机微分方程组系统来研究云计算中资源分配过程的建模。这种新的随机模型将通过分析大量的非平稳云数据来进行模型识别。由于控制不断增长的云计算基础设施所需的人工工作量极高且效率低下，使得云无法管理，因此我们将设计一个基于新的随机差分模型的自主计算环境、协议和控制算法来控制云资源。该系统将实施自我优化、自我配置和自我学习算法，模拟人工操作员的决策和与云的交互。用于云资源控制的自主计算将对云资源进行实时优化管理，降低计算能力和能源消耗。我们建议研究复杂自主计算系统的设计，实现云计算自治的智能、局部和全局控制器。我们还将为云控制器提供自适应算法，以实施自组织策略，以优化虚拟服务器的使用。自我感知功能将提供有关资源状态及其链接的资源的信息。一个轻型原型将在申请者实验室托管的云上实施。软件定义网络和软件定义服务的最新进展将用于原型实施。

The objectives of this research application are to investigate selected theoretical and experimental topics related to the design and implementation of specific components of the next-generation of Information Technology Infrastructures for supervisory control of enterprise processes. It is common knowledge that Information and Communication Technologies are the central piece of the new digital economy due to the commoditization of computer and networking technologies. Recently, the World Economic Forum predicted that the new digital economy will be the major industrial component in any developed country. More and more, applications and data are migrating away from customer premises and into cloud environments. Reports show that an average Cloud Computing infrastructure consumes the same amount of energy like 25,000 households. The advancements made in past years in the area of distributed and cloud computing have led to the production of massive software technologies, hosted by large and extremely costly data centres, while the “green computing” movement promotes improving cloud efficiency in regards to the energy consumed. Despite sustained research efforts, a formal model of the dynamics of the energy consumption in Cloud Computing infrastructures, due to the computing, storage and networking resource utilization, is still lacking. We propose to investigate the modeling of resource allocation processes in Cloud Computing by Stochastic Differential Equations systems. This new stochastic model will be devised by analyzing large amounts of non-stationary cloud data for model identification. Because the manual effort needed to control a growing cloud computing infrastructure is extremely high and inefficient, making the Cloud unmanageable, we will devise an Autonomic Computing environment, protocols and control algorithms based on the new Stochastic Differential model to control cloud resources. This system will implement self-optimization, self-provisioning, and self-learning algorithms, which mimic human operator decisions and interactions with the cloud. The Autonomic Computing for cloud resource control will manage optimally the cloud resources in real-time reducing the computational power and the energy consumption. We propose to investigate the devising of complex Autonomic Computing system implementing intelligent, local and global controllers for clouds computing self-governance. We will also produce adaptive algorithms for the cloud controller to implement self-organizing strategies in order to optimize virtual servers' usage. A self-awareness feature will provide information about the state of resources and the resources it links to. A light prototype will be implemented on the cloud hosted in the applicant's laboratory. Recent advances on Software Defined Networking and Software Defined Services will be used for prototype implementation.