Evolutionary computing approach to optimize superframe scheduling on industrial wireless sensor networks

Abstract

The industrial wireless sensor domain has undergone a shift in paradigm as a consequence of Internet of Things (IoT), a thriving technology that has been leading the way in short-range and fixed wireless sensing. One of the problems associated with industrial wireless sensor networks (IWSNs) is finding the optimal solution for minimizing defect time in superframe scheduling. This paper proposed a method based on the use of evolutionary algorithms, namely particle swarm optimization (PSO), orthogonal learning PSO, genetic algorithm (GA), and modified GA for optimizing the superframe scheduling. Additionally, we evaluated a contemporary method, deadline monotonic scheduling, on the ISA 100.11a protocol. The use of this standard as a case study means that the presented 72 simulations are object-oriented, with numerous variations in the number of timeslots and wireless sensor nodes. The simulation results show that the use of GA and modified GA can improve the performance in terms of idle, missed deadlines, memory consumption, and processing time comparing to other metaheuristic algorithms. A comprehensive analysis and detailed performance evaluation are provided in the paper.

Keywords

• Evolutionary programming
• Genetic algorithm
• Industrial wireless sensor network (IWSN)
• Internet of Things (IoT)
• ISA 100.11a
• Superframe scheduling