Affiliations: Embedded Systems Laboratory, School of Computing, SASTRA Deemed University, India
Correspondence:
[*]
Corresponding author. A. Umamakeswari, Embedded Systems Laboratory, School of Computing, SASTRA Deemed Univer-sity, 613401, India. E-mail: r.ezhilwin@gmail.com.
Abstract: Edge computing offers potential benefits to applications working in IoT (Internet of Things) and CPS (Cyber Physical Systems) environments by bringing the power of computing proximate to the devices, which demand high computational resources. As computational capabilities are currently untapped in edge devices like the IoT gateway, the computational intensive part of an application like a thread, a module or a task can be offloaded to the edge devices rather than to the cloud by the end devices. In this paper, an approach that employs Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) is used to determine the near optimal solution for scheduling offloadable components in an application, with the intent of significantly reducing the execution time of an application and energy consumption of the smart devices. With a new inertial weight equation, an Adaptive Genetic Algorithm – Particle Swarm Optimization (AGA-PSO) algorithm is proposed which uses GA’s ability in exploration and PSO’s ability in exploitation to make the offloading optimized without violating the deadline constraint of an application.
Keywords: Internet of things (IoT), edge computing, computation offloading, application partitioning, particle swarmoptimization, genetic algorithm
