Development of a novel Quality of Service and energy balancing control engine to reduce and measure the carbon emissions of distributed computing networks for high throughput computing.

慈善引擎（CE）平台（由Worldwide Computing Company Ltd运营）是一个网格计算解决方案，利用剩余的“空闲时间”计算能力为全球企业和学术客户提供计算服务，以满足高性能计算（HPC）的需求，例如科学模拟或统计分析。企业用户为服务付费，而慈善机构或研究机构则免费获得额外的计算能力作为慈善捐款。测量和报告能源需求和随后的碳排放，同时保持足够的服务质量（QoS）以满足客户的计算需求，并使企业能够履行企业社会责任报告要求。该项目的目的是通过分配任务，大幅减少CE基础设施的能源消耗和二氧化碳排放以最小化每个任务消耗的能量的量的方式，同时满足终端用户的QoS约束。通过降低电力成本和减少二氧化碳影响，将与基础设施所有者分享收益，并通过减少其运营的二氧化碳影响与最终用户分享收益。该项目将利用认知网络有限公司的研究所取得的技术进步，认知网络有限公司已经设计和实现了几个概念验证的能源/该联盟将开发一个基于强化学习的强大的QoS能量/平衡引擎，该引擎将集成到CE分布式网络中，基于实时QoS（包括网络延迟、服务器响应时间和服务器能耗）来最优地选择服务器节点。

The Charity Engine (CE) platform (ran by Worldwide Computing Company Ltd) is a grid computing solution that harnesses surplus 'idle time' computing power to provide computing services to both enterprise and academic customers worldwide to meet the high-performance computing (HPC) needs such as scientific simulations or statistical analyses. Corporate users pay for the service, while charitable or research institutions receive the excess computing power free of charge as a charitable contribution.There is a need for a grid computing solution that reduces, measures and reports both energy requirements and subsequent carbon emissions whilst maintaining an adequate Quality of Service (QoS) to meet customer compute demands and enables corporates to fulfil CSR reporting requirements.The aim of this project is to substantially reduce the energy consumption and CO2 emissions of the CE infrastructure by allocating tasks in a manner which minimizes the amount of energy consumed per task, while meeting the QoS constraints of the end users. The benefit will be shared with the owners of the infrastructure through reduced electricity costs and CO2 impact, and with the end users through reduced CO2 impact of their operations.To achieve this aim, this project will exploit the technical advancements achieved by research at Cognitive Networks Limited which has designed and implemented several proof-of-concept Energy/QoS balancing control systems in several target testbeds.The consortium will develop a robust QoS energy/balancing engine based on Reinforcement Learning that which will be integrated into the CE distributed network, to optimally select server nodes based on real-time QoS including network delay, server response time and server energy consumption.