Multi-objective cost-aware bag-of-tasks scheduling optimization model for IoT applications running on heterogeneous fog environment

Abstract

Recently introduced by CISCO, the fog computing paradigm extends cloud-based computing services near the Internet of Things (IoT) devices to serve their applications, aiming to increase performance. This new computing paradigm tries to meet the requirements of emerging IoT applications, which mostly are geographically distributed and location-aware, low-cost, and delay-sensitive. One of the frequently applied applications is a bag-of-tasks (BoT) of IoT applications, each of which includes several independent tasks. However, inefficient task scheduling approaches for parallel executions may degrade the delivered quality of service (QoS) to users and also increase the providers’ total cost of ownership (TCO). The reliable execution commensurate with users’ service level agreement (SLA) and affordable services tailored to their budget at the same time is favorable for both users and fog service providers because the highly reliable execution is a reputational metric that increases the users’ trust and adherence. To this end, this paper formulates the task scheduling for the execution of BoT in IoT applications to a multi-objective optimization problem with execution time, costs and reliability perspectives. To measure total costs, both the execution cost and monetary cost models are presented. To measure the resource reliability, the new scheduling failure factor (SFF) metric is also presented, which is a kind of cost parameter. To solve this combinatorial NP-Hard problem, a multi-objective cost-aware discrete grey wolf optimization-based algorithm (MoDGWA) is proposed. To verify the performance of MoDGWA, diverse scenarios were conducted and the performance of the proposed algorithm was compared against other state-of-the-arts. The extensive simulation results prove the average dominance of MoDGWA, and it has 0.55 %, 7.28 %, 10.20 %, and 45.83 % reduction against other state-of-the-art in terms of makespan, ToC, TSFF, and the cost score respectively. Note that, the implementation of the MoDGWA is available on https://github.com/mirsaeid123/Mirsaeid.