Next Generation of IoT Backend Middleware for Edge and Cloud

物联网(IoT)、移动设备和传感器网络的出现预计将导致分布式数据源的数量呈指数级增长，预计到2021年，连接设备将达到270亿台。然而，当前的移动网络不能支持需要低延迟的下一代应用，或者产生可能使网络基础设施不堪重负的大量数据的下一代应用。例如，智能个人助理、医疗患者监控和智能交通系统。由于同样的低延迟要求，在广域云上使用服务器也不是一种选择。边缘计算是一种新兴的替代方案，它使用附近的资源以及通往云的路径上的节点，在非常多的地理分布的站点上提供计算和存储。在创建点附近处理数据有几个好处：它减少了广域网络带宽需求，降低了计算依赖于本地数据的结果的延迟，并有助于限制敏感数据传播的地理区域。尽管前景看好，但开发边缘计算应用程序是具有挑战性的：通信链路缓慢或不可靠，工作负载和网络不断变化，边缘设备经常动力不足；应用程序部署和共享数据存储也很困难。该项目旨在通过构建一个统一的模型来开发和部署同时在边缘和云上运行的应用程序来弥合这一差距。它探索了在分层边缘计算设置上进行数据处理的新愿景，该设置由位于网络边缘和传统广域云之间的具有不同功能的数据中心组成。该项目有三个主要研究重点：(1)通用功能即服务运行时和部署环境；(2)共享分布式存储系统；以及(3)大数据处理和分析框架。

The advent of the Internet-of-Things (IoT), mobile devices, and sensor networks is expected to lead to an exponential increase in the number of distributed data sources, with a predicted 27 billion connected devices by 2021. Current mobile networks, however, are not able to support next generation applications that require low latency, or that produce large volumes of data that can overwhelm the network infrastructure. Examples include intelligent personal assistants, medical patient monitoring, and intelligent transportation systems. The use of servers on the wide-area cloud is also not an option due to the same low-latency requirements. Edge computing is an emerging alternative that uses nearby resources, as well as nodes on the path to the cloud, to provide computation and storage over very large number of geographically distributed sites. Processing data close to the point of creation has several benefits: it reduces wide-area network bandwidth demands, it lowers the latency to compute results that depend on local data, and it helps limit the geographic area where sensitive data propagates. Despite their promise, developing edge computing applications is challenging: communication links are slow or unreliable, workload and network are ever-changing, and edge devices are often under-powered; application deployment and shared data storage are also difficult.This project aims to bridge this gap by building a unified model for developing and deploying applications that run simultaneously on the edge and the cloud. It explores a new vision for data processing on a hierarchical edge computing setup consisting on data centers with varying capabilities that are interposed between the edge of the network and the traditional wide area cloud. The project has three main research thrusts: (1) a common Function-as-a-Service runtime and deployment environment; (2) a shared distributed storage system; and (3) a framework for big data processing and analytics.