EdgeSphere: A Three-Tier Architecture for Cognitive Edge Computing

在物联网设备在边缘产生海量数据并在将这些数据传输到云端进行分析方面面临挑战的情况下，边缘计算变得越来越重要。另一方面，如果能够克服设备能力、连接性和异构性等挑战，利用边缘数据对于提供认知应用程序至关重要。本文提出了一种新型的三层架构，称为EdgeSphere，利用边缘设备的资源在边缘进行数据分析。与现有的云和移动应用程序不同，EdgeSphere应用程序跨越云、边缘网关和边缘设备。在其核心，EdgeSphere基于Apache Mesos来优化资源使用和调度。EdgeSphere已应用于实际场景，本文描述了面临的工程挑战以及创新解决方案。

Edge computing is becoming increasingly important due to the challenges posed by transporting massive amounts of data generated by IoT devices to the cloud for analysis. This paper proposes a new architecture, EdgeSphere, to analyze data at the edge and overcome challenges such as device capabilities, connectivity, and heterogeneity.

EdgeSphere is a three-tier architecture that optimizes resource usage and scheduling using Apache Mesos. It spans across cloud, edge gateways, and edge devices to offer cognitive applications.

The paper describes the engineering challenges faced and innovative solutions for practical scenarios. The EdgeSphere architecture has been applied to real-world scenarios. The paper also discusses the benefits of using EdgeSphere for cognitive applications and the potential for future research in this area.

Related work includes research on edge computing, IoT, and cloud computing such as 'Mobile Edge Computing, Fog et al.: A Survey and Analysis of Security Threats and Challenges' and 'The Role of Edge Computing and the IoT in the Digital Transformation of Business'.