Resource Allocation and Scheduling of Real-Time Workflow Applications in an IoT-Fog-Cloud Environment

Abstract

The explosive growth of the Internet of Things (IoT) has led to the emergence of the IoT-fog-cloud continuum, in an attempt to facilitate the real-time processing of IoT data. In such multi-tier environments, it is crucial to adopt an efficient resource allocation and scheduling scheme, in order to provide effective load balancing and timeliness for the real-time workload. A load balancing approach that has been proven to be efficient and effective in traditional distributed environments, is the power of two choices – or $d$ choices, in its general form. Only recently has this technique been examined in multi-tier environments, without considering, however, important aspects of such frameworks. To this end, in this paper we propose and investigate three resource allocation and scheduling heuristics for real-time workflow jobs in an IoT-fog-cloud environment. The first strategy, performs exhaustive search at each scheduling step in order to find the most suitable resource in the fog and cloud layers for the workload assignment. On the other hand, the two other policies adopt the power of two choices approach. The simulation results shed light on interesting insights regarding the performance and applicability of each method.