CSR: Medium: Collaborative Research: A Data-Centric Architecture for Pervasive Edge Computing in Heterogeneous Extensible Distributed Systems

该项目旨在开发一种新的以数据为中心的架构和相应的协议/算法，以实现基于对等的数据发现、传播和处理，用于异构边缘设备（例如，智能手机、平板电脑），而不像目前那样依赖于任何集中式后端。该项目使用命名数据来实现大量异构边缘设备之间的直接数据共享和处理。该研究系统地研究：1）适合于精确引用和表征各种数据的结构和构造; 2）数据发现和名称匹配机制，用于设备查找可用数据并获取所需数据; 3）透明处理，从原始传感样本中导出和提取新数据。该项目结合了架构、协议和算法设计、优化、测试床实现和软件模拟，对上述问题进行了全面研究，这项研究将使下一代移动的传感应用不再需要集中的后端，从而完全消除相关的金钱、人力成本，并对互联网连接的可用性和性能的依赖。PI正在开发一个新的研究生课程，培训研究生，并接触到本科生和高中生，以建立未来的应用程序。与领先的研究和开发机构进行联合探索，以应用研究成果。出版物，软件和实验数据与社区共享，以促进对该架构的进一步研究。

The project aims to develop a novel data-centric architecture and accompanying protocols/algorithms to enable peer based data discovery, dissemination and processing for future mobile sensing applications among heterogeneous edge devices (e.g., smartphones, tablets) without relying on any centralized backend as is currently done. The project uses named data to enable direct data sharing and processing among vast numbers of heterogeneous edge devices. The research systematically investigates: 1) Naming structures and constructs suitable for precise reference and characterization of various kinds of data; 2) Data discovery and name matching mechanisms for devices to find what data are available and acquire desired ones; 3) Transparent processing that derives and extracts new data from raw sensing samples. The project combines architecture, protocol and algorithm design, optimization, testbed implementations and software simulations for a full-scale study on the above issues.This research will enable a future generation of mobile sensing applications no longer requiring a centralized backend, thus completely eliminating the associated monetary, manpower costs, and dependence on the availability and performance of Internet connectivity. The PIs are developing a new graduate course, training graduate students, and reaching out to undergraduate and high school students to build futuristic applications. Joint explorations with leading research and development organizations are conducted to apply the research results. The publications, software and experimental data are shared with the community to foster further research on this architecture.