| Abstrakt: |
In the world of the growing internet of things (IoT), multi-access edge computing (MEC) is an important innovation. It strategically places servers in decentralized areas to reduce bandwidth usage and transmission latency. This paper focuses on the essential initial phase of MEC deployment, focusing on the major challenge of optimizing the placement and allocation of edge servers. Such optimization is central to increasing the benefits of edge computing systems while reducing costs and energy consumption and improving quality of service (QoS). Traditional optimization methods often treat variable categories homogeneously and neglect directional search considerations, thus compromising search efficiency for large-scale problems. To address these issues, our study develops a complex set of methods that combine decision variable classification and elitism to improve the optimization path and lead to a satisfactory solution. This innovative approach pursues two objectives: First, it articulates an objective to minimize energy consumption, incorporating considerations of proportional load balancing, latency reduction, and fair distribution of users across edge servers. Second, this objective is combined with the latency reduction objective, resulting in a dual-topic optimization task. We compare the proposed method with well-known approaches such as non-dominated sorting genetic algorithm II (NSGA-II), non-dominated sorting genetic algorithm III (NSGA-III), and multi-objective particle swarm optimization (MOPSO). Our findings demonstrate a minimum 3.2% reduction in latency in homogeneous server environments and a remarkable 5.02% reduction in heterogeneous server environments. Energy consumption also drops significantly compared to similar benchmarks. Energy consumption has also fallen significantly compared to comparative benchmarks. The application of the non-parametric Friedman test demonstrates the superiority of the proposed method, which ranks highest among state-of-the-art methods and emphasizes statistical precision. Through a detailed technical discussion, the paper substantiates the feasibility and superiority of the proposed optimization method in addressing the complex requirements of edge server placement and allocation within the MEC paradigm. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink