A Hybrid Latency- and Power-Aware Approach for Beyond Fifth-Generation Internet-of-Things Edge Systems

Abstract

Fifth-generation (5G) empowered internet of things (IoT) edge networks suffer from latency in delay-sensitive applications. To fulfil the low latency requirements of beyond fifth-generation (B5G)-IoT applications and provide quality of service (QoS) to IoT-edge communication, it is important to minimize server delay. Furthermore, 5G-IoT systems consume more power than their predecessors, which is a concern given the growing size of future IoT networks. This research presents a hybrid latency and power-aware approach for B5G-IoT networks (HLPA B5G-IoT) that minimizes latency with minimum overhead on battery-constrained IoT nodes while simultaneously providing a power-efficient solution for B5G-IoT-edge networks. HLPA B5G-IoT has a novel algorithm classifier tool (ACT) for selecting appropriate optimization algorithms based on the characteristics and requirements of B5G-IoT systems. The ACT matrix not only parametrically compares HLPA B5G-IoT with existing approaches but also identifies crucial parameters that enable algorithm selection for load balancing and energy efficiency. In this paper, metaheuristic algorithms, i.e., biogeography-based optimization (BBO) and grey wolf optimization (GWO), are tailored to meet the requirements of load balancing and power efficiency in IoT-edge systems. The proposed load-balancing algorithm reduces latency and improves overall network performance by 33.33%, 27.45%, 23.52%, 21.56%, 13.72%, 11.76%, and 7.84% compared with simulated annealing (SA), genetic algorithm (GA), particle swarm optimization (PSO), bacteria foraging algorithm (BFA), ant colony optimization (ACO), bat algorithm (BA), and genetic SA PSO (GSP), respectively. The power-efficiency algorithm consumes 46.6%, 40%, 32.2%, 27.7%, 15.5%, 11.1%, and 6.6% less energy compared with SA, GA, PSO, BFA, ACO, BA, and GSP, respectively.