Fog computing middleware for distributed cooperative data analytics

Abstract

This paper presents an innovative fog computing middleware for distributed cooperative data analytics (DCDA) in the Internet of Things (IoT). The existing IoT systems gather data in a central place (e.g., cloud) for post-processing and high-level situation awareness. Such a collecte-and-compute paradigm cannot sustain the exponential sensors and data growth — the communication bandwidth is limited by spectrum and cannot catch up the data growth rate. In addition, many critical cyber-physical applications demand time-sensitive decisions, and privacy concerns prefer to avoid data collection. Distributed cooperative data analytics for high-level situation awareness can avoid costly data collection and meet the time and privacy requirements, and has received much attention in the past several years. In this work, we propose a DCDA architecture and middleware to support flexible and scalable DCDA, and test and validate the proposed middleware in seismic and ambient noise imaging case studies. The evaluation results in a fog computing testbed demonstrate that the DCDA middleware allows scalable and fault-tolerant data analytics and is suitable for real-time situation awareness under the bandwidth and communication constraints.