CICI: UCSS: Trusted Resource Allocation in Volunteer Edge-Cloud Computing Workflows

网络智能边缘设备（例如科学仪器、边缘服务器、传感器）和相关数据源的空前增长导致科学应用社区出现数据洪流。数据处理越来越依赖机器学习算法和云平台来应对数据的异构性、规模和速度。同时，还有大量低成本计算资源可用于“志愿者边缘云计算”（VEC），社区中具有不同背景的合作者可以贡献自己的资源，形成分布式基础设施来执行科学工作流程。 VEC 范式有可能使不同学科（例如地理学、神经科学、制造、农业）的许多科学工作流程社区受益。该 CICI 项目调查了由于协作同行之间在 VEC 节点的预期性能、安全性以及工作流程策略方面缺乏信任而导致更广泛采用 VEC 的障碍。该项目的目标是开发 VECTrust 架构和相关框架，以支持 VEC 环境中科学数据密集型工作流程的可信资源分配。在应用程序社区对 KBCommons 科学网关中用于多组学数据集成的 VEC 的需求的推动下，正在开发威胁模型，以告知协作伙伴的资源分配选项的安全性。此外，正在开发一种信任模型，该模型使用两阶段概率模型来使用与工作流执行因素相关的指标来定义 VEC 集群资源的可信度，这些因素包括：性能、敏捷性、成本和安全性。通过可用性研究，正在研究一种规范性算法，以确保可能缺乏保护工作流程的知识、背景或资源的领域科学家可以轻松、无缝地采用 VEC 集群。项目外展活动的重点是创建和传播开源和协作信任的 VEC 资源编排工具、数据集和教育材料，以培训下一代人才。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

The unprecedented growth in networked smart edge devices (e.g., scientific instruments, edge servers, sensors) and related data sources has caused a data deluge in scientific application communities. The data processing is increasingly relying on machine learning algorithms and cloud platforms to cope with the heterogeneity, scale, and velocity of the data. At the same time, there is an abundance of low-cost computation resources that can be used for “volunteer edge-cloud computing” (VEC), where collaborators in a community with various backgrounds contribute their resources to form a distributed infrastructure to execute scientific workflows. The VEC paradigm has the potential to benefit many scientific workflow communities in diverse disciplines (e.g., geography, neuroscience, manufacturing, agriculture). This CICI project investigates the barriers for wider adoption of VEC caused due to lack of trust amongst the collaborating peers in terms of expected performance, and security of the VEC nodes, as well as workflow policies. This project’s goal is to develop a VECTrust architecture and a related framework for support of trusted resource allocation in VEC environments for scientific data-intensive workflows. Driven by application community requirements for VEC in the KBCommons science gateway for multi-omics data integration, a threat model is being developed to inform the security of the resource allocation options for the collaborating peers. In addition, a trust model is being developed that uses a two-stage probabilistic model to define trustworthiness of VEC cluster resources using metrics relevant to workflow execution factors that include: performance, agility, cost and security. Through usability studies, a prescriptive algorithm is being investigated to ensure that domain scientists who may lack the knowledge, background or resources to secure workflows can easily and seamlessly adopt VEC clusters. Project outreach activities are focused on creation and dissemination of open-source and collaborative trust-enabling VEC resource orchestration tools, datasets and educational materials to train next-generation talent.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.