This paper presents a processing model for big IoT data. The model includes a continuous delivery scheme based on building blocks for constructing software pipelines from the edge to the cloud. It also includes a data preparation scheme based on parallel patterns for establishing, in an efficient manner, controls over the production and consumption of IoT data. This scheme adds data properties such as cost-efficiency storage, security, and reliability, which are useful to avoid alterations in data and repudiation situations as well as to mitigate risks still arisen in the cloud such as confidentiality violations and service outages. An overlay structure, including planes such as pub/sub, control, and preservation, integrates the proposed schemes into software pipelines. The proposed model was developed in both prototype and simulator of software pipelines. Case studies were conducted based on pipeline services deployed from the edge, passing from the fog to the cloud for processing and managing real climate data repositories, which were produced by three different data sensor sources, such as ground stations deployed on Mexico and Spain, as well as small distributed IoT devices. Information sharing patterns for end-users to retrieve raw and/or processed IoT data were also studied. The experimental evaluation revealed the feasibility of using continuous delivery scheme to create dataflows from the edge to the cloud, the efficacy of the overlay structure to create information sharing patterns, as well as the efficiency of data preparation schemes and parallel patterns to improve the end-user service experience in comparison with traditional state-of-the-art solutions.

本文提出了大物联网数据处理模型。该模型包括基于构建块的连续交付方案，用于构建从边缘到云的软件管道。它还包括基于并行模式的数据准备方案，以有效地建立对物联网数据生产和使用的控制。此方案添加了诸如成本效益存储，安全性和可靠性之类的数据属性，这些属性可用于避免更改数据和抵赖情况，以及减轻仍然存在于云中的风险，例如违反机密性和服务中断。包括平面（例如发布/订阅，控制和保存）的覆盖结构将提出的方案集成到软件管道中。该模型是在软件管道的原型和仿真器中开发的。案例研究是基于从边缘部署的管道服务（从雾到云）进行处理和管理实际气候数据存储库的，这些数据存储库是由三种不同的数据传感器源（例如部署在墨西哥和西班牙的地面站）生成的。以及小型分布式物联网设备。还研究了最终用户检索原始和/或处理后的物联网数据的信息共享模式。实验评估表明，使用连续交付方案来创建从边缘到云的数据流的可行性，使用覆盖结构创建信息共享模式的有效性，