Energy-Aware Service Allocation: A Crow Search-Based Approach

Abstract

The allocation of services from the cloud to edge servers in the Industrial Internet of Things (IIoT) emerges as a promising solution for providing time-critical services by shifting computation near sensor nodes while improving the energy cost of the cloud servers and energy consumption of sensor nodes. However, allocation of services faces challenges due to– a) limited computation, storage, and energy constraint of edge servers, b) increasing brown energy consumption of cloud servers, and c) compatibility of communication protocols used by edge servers. In this work, we propose crow search-based two schemes, EAllocS and EAllocM, for allocating unsplittable and splittable services to edge servers. In contrast to existing literature, we consider energy consumption, delay, utility score of edge servers, compatibility of the communication protocols, and energy cost of the cloud servers. Extensive simulation results show that EAllocS reduces energy consumption, delay, and maximum load by 56.7%, 49.9%, and 41.4% compared to the benchmarks. EAllocM reduces energy consumption, delay, and maximum load by 58.5%, 24.8%, and 38%, respectively. The results show that proposed schemes save more energy by reducing the energy consumption of processing services. EAllocS and EAllocM also improve total device allocation cost compared to the benchmark schemes.